test(oracle): compare reading citations against missalemeum

Layer 4 (test_oracle.ml) now extracts and compares the Epistle/Gospel
citation for every one of the 730 days in the 2026-2027 fixture, the
same axis Task 8 already added to layer 3. tools/
extract_missalemeum_oracle.py gained a structural citation-span
detector (the first "*...*"-wrapped span under 40 chars containing a
chapter:verse-shaped digit pair), verified exhaustively over all 1 458
Lectio/Evangelium sections. 728 of 730 days are comparable; the 2 that
are not (Good Friday, both years) have no such section in missalemeum's
own JSON at all -- a structural limit on the oracle's own side (M23),
not a colitur gap. 451 of 730 days differ and are all explained across
19 cited entries; 0 unexplained. Chain step 4 (the Commons) has no
layer-4 witness: confirmed empirically (a scratch sweep, not merely
re-asserted from Task 6's own report) that none of the 15 Commons/proper
saints is ever the observed office in this window.

The comparison surfaced three genuine colitur bugs in
data/ef/lectionary.sexp, all fixed at the source in
tools/bootstrap_lectionary.ml:

1. Holy Name of Jesus (ef-holy-name-sunday/ef-holy-name) carried a
   mis-derived citation, borrowed by analogy from an unrelated ini
   section ("ef-christmas-sunday-0", the Sunday-within-the-Octave's own
   Mass) on the wrong assumption the two share a Mass. Disproved by
   missalemeum's own extraction for both real Holy Name Sundays in its
   window, confirmed against the Missal itself, twice, both scans
   (Acts 4:8-12/Luke 2:21, not Gal 4:1-7/Luke 2:33-40).
2. RG 17(a)'s own "diebus ferialibus" rubric (immediately following the
   Holy Name heading, both scans) was never implemented: 2-5 January
   ferias repeat 1 January's Mass, 7-12 January ferias before Holy
   Family Sunday repeat Epiphany's, and Holy Family's own week has a
   named formula ("Missa Dominicae Primae post Epiphaniam"). All three
   fell to step 3's undifferentiated Sunday walkback instead. Fixed for
   three 5-weekday families, deliberately excluding Saturday from all
   three -- RG 78's BVM Saturday Office wins there unconditionally,
   confirmed against missalemeum's own title on this window's own such
   Saturdays.
3. Corpus Christi and Sacred Heart (both real, never-impeded I-class
   movable feasts) had no lectionary entry at all. Both engines
   independently fell back to Trinity Sunday's own citation and
   silently agreed on the wrong answer -- the "Holy Thursday violet in
   both" shape CLAUDE.md names for exactly this reason.

Because lectionary.sexp is shared with layer 3 (the lectio differential,
2005-2050), these fixes changed colitur's output for the whole domain,
not only 2026-2027. data/ef/expected-divergences.sexp and
test_differential.ml needed matching updates: C1 narrowed 361->173 and
C25 narrowed 57->13 (rows that now resolve correctly moved to new
entries), C16's note corrected (count unchanged), and six new entries
added (C29-C34, 397 rows total, all verdict colitur). test_lectionary.ml
and test_lectionary_ef.ml's own pins were updated to match (entry count
131->148; one step-3 regression test re-pointed to a date that still
genuinely exercises that step).

One entry, M26 (30 rows, verdict open), is a genuine unfixed gap: colitur's
step-3 "resume the preceding Sunday" fallback is its only ferial-Mass
mechanism, and missalemeum's own selection is more granular in three
confirmed ways (a rotating votive Mass of the BVM, a named weekly
formula for at least one week, and later Time-after-Pentecost weeks this
task's own investigation did not narrow to a single rule) -- the same
"needs a season-keyed selection mechanism, new behaviour not new data"
shape data/ef/expected-divergences.sexp's own C6/C25 and Task 6's own
residual concern already name, out of this task's own scope to build.

dune test: 358 tests green. COLITUR_EXHAUSTIVE_SWEEP=1 dune test
--force: 359 tests green, 89.5s.

1 files changed, 154 insertions, 4 deletions

diff --git a/test/test_differential.ml b/test/test_differential.ml
index 655a054..a89b17b 100644
--- a/test/test_differential.ml
+++ b/test/test_differential.ml
@@ -699,6 +699,57 @@ let epiphany_week_shift_years_2005_2050 = [ 2008; 2013; 2019; 2030; 2036; 2041;
 
 let year_of_date date = int_of_string (String.sub date 0 4)
 
+(* C29-C34 -- Task 9 (branch ef-lectionary, layer-4 oracle): six colitur-side
+   citation fixes, all found via the SAME mechanism -- missalemeum (test/
+   test_oracle.ml, 2026-2027) disagreeing with colitur's OLD step-3
+   fallback/absent-entry answer, traced to an explicit primary-source
+   citation each time (tools/bootstrap_lectionary.ml's own
+   [holy_name_entries]/[epiphanytide_opening_entries]/
+   [movable_feast_entries] carry the full scan citations, not repeated
+   here). lectio has no equivalent data for any of the six (its own ini
+   either has no entry, matching colitur's OWN prior gap for
+   [ef-corpus-christi]/[ef-sacred-heart], or the SAME mis-borrowed value
+   colitur used to carry for the January families before this task's own
+   fix, see [colitur_keys]'s own CORRECTED note in that generator) -- so
+   every occurrence across the WHOLE 2005-2050 differential window newly
+   diverges, not only the 2026-2027 oracle window that found it. Each is
+   its own id, not folded into a neighbour, per this file's own standing
+   discipline (C14/C15/C16 above): a slug family's own citation fix is a
+   distinct fact from another family's, even where the underlying task
+   and citation shape are similar. *)
+let holy_name_sunday_slug = "ef-holy-name-sunday"
+
+let christmas_1_weekday_slugs =
+  List.map (fun wd -> "ef-christmas-1-" ^ wd) [ "monday"; "tuesday"; "wednesday"; "thursday"; "friday" ]
+
+let christmas_2_weekday_slugs =
+  List.map (fun wd -> "ef-christmas-2-" ^ wd) [ "monday"; "tuesday"; "wednesday"; "thursday"; "friday" ]
+
+let time_after_epiphany_1_weekday_slugs =
+  List.map
+    (fun wd -> "ef-time-after-epiphany-1-" ^ wd)
+    [ "monday"; "tuesday"; "wednesday"; "thursday"; "friday" ]
+
+(* C31/C32's OWN Saturday ripple, checked for and confirmed real, not
+   theorised: [ef-christmas-1-saturday] and [ef-time-after-epiphany-1-
+   saturday]/[ef-christmas-2-saturday] deliberately have NO direct entry
+   (RG 78's BVM Saturday Office, above) -- but a Saturday still falls
+   through to step 3's OWN preceding-Sunday walkback, and on the (large
+   minority of) years where that walk lands on [ef-holy-name-sunday]
+   specifically, it now inherits THAT slug's newly-corrected citation as
+   a side effect, with no direct entry of its own involved at all.
+   [ef-christmas-1-saturday]'s own walkback structurally can never reach
+   that far forward (it only ever lands in December, C25's own unchanged
+   population, confirmed by the pinned test_lectionary_ef.ml case) --
+   only the OTHER two families' Saturdays are affected, so only they need
+   the wider slug list; a Saturday whose value is genuinely unaffected
+   this year simply produces no row needing an id at all ([diffs = []]),
+   not a false positive. *)
+let christmas_2_weekday_slugs_with_saturday = christmas_2_weekday_slugs @ [ "ef-christmas-2-saturday" ]
+
+let time_after_epiphany_1_weekday_slugs_with_saturday =
+  time_after_epiphany_1_weekday_slugs @ [ "ef-time-after-epiphany-1-saturday" ]
+
 (* C20 (task 8): lectio genuinely computes NO reading at all -- "-"/"-" on
    both parts -- for 14 civil days across 2005-2050, all seven of the
    readingless class-3 saints Task 6 gave a proper or a Common
@@ -950,10 +1001,28 @@ let september_ember_sat_slug = "ef-september-ember-sat"
    respectively" (the right four numbers, the wrong weekday each was
    attached to). Re-measured directly against the real resolver:
    Wednesday 6, Thursday 19, Friday 19, Saturday 13 (6 + 19 + 19 + 13 =
-   57, no residue). *)
-let rg69_one_hop_slugs =
-  [ "ef-christmas-1-wednesday"; "ef-christmas-1-thursday"; "ef-christmas-1-friday";
-    "ef-christmas-1-saturday" ]
+   57, no residue).
+
+   NARROWED, Task 9 (branch ef-lectionary, layer-4 oracle): three of the
+   four weekdays (Wednesday/Thursday/Friday) now have their OWN direct
+   lectionary entry (tools/bootstrap_lectionary.ml's own
+   [epiphanytide_opening_entries], RG 17(a)'s "diebus ferialibus a 2 ad 5
+   ianuarii Missa dicitur ut die 1 ianuarii", both scans) -- they no
+   longer reach step 3's one-hop walkback THIS entry's own citation
+   describes at all, so keeping them here would misdescribe what
+   actually happens on those three weekdays now. See [C30] below for
+   their own (DIFFERENT, larger) new population -- colitur is RIGHT
+   there, not merely differently wrong, so it is a separate id with its
+   own [verdict colitur], not a re-adjudication of this one. Only
+   [ef-christmas-1-saturday] remains here: deliberately NOT given a
+   direct entry (RG 78's BVM Saturday Office wins that day
+   unconditionally whenever nothing else does, {!Rite_ef.Temporal_ef}'s
+   own [bvm_saturday_names] -- giving it the "repeat 1 January" citation
+   would be wrong on exactly the days this entry's own citation already
+   describes as correct for the OTHER four weekdays), so this ONE
+   weekday's own step-3 walkback -- and this entry's own citation,
+   "Tit. 3, 4-7" -- is genuinely unchanged by this task. *)
+let rg69_one_hop_slugs = [ "ef-christmas-1-saturday" ]
 
 (* [layer_c_reason l c diffs] returns the [data/ef/expected-divergences.sexp]
    [id] this row-pair's remaining (post Layer A/B) diff set belongs to, or
@@ -1046,6 +1115,48 @@ let layer_c_reason (l : row) (c : row) diffs =
     && subset diffs [ Slug_f; Rank; First_f; Gospel_f ]
     && String.equal c.slug "ef-holy-name"
   then Some "C16"
+  else if
+    (* C29 -- Task 9: the Holy-Name-SUNDAY shape (RG 17(a)'s own primary
+       Sunday case, distinct from C16's fallback above) -- Season/Slug_f
+       already agree (Layer A already normalises lectio's own
+       "ef-christmas-sunday-0" for this civil date onto colitur's
+       "ef-holy-name-sunday", the SAME alias this generator's OWN
+       [colitur_keys] used to (wrongly) borrow a citation through --
+       Layer A's alias and the generator's OLD citation-sourcing bug were
+       two separate uses of the same underlying fact, only one of which
+       was ever wrong), so only the citation itself is new. *)
+    diffs = [ First_f; Gospel_f ] && String.equal c.slug holy_name_sunday_slug
+  then Some "C29"
+  else if
+    (* C30 -- Task 9: RG 17(a)'s own "diebus ferialibus a 2 ad 5 ianuarii"
+       rubric, the five non-Saturday weekdays (Saturday stays C25's own,
+       narrowed note above explains why). *)
+    diffs = [ First_f; Gospel_f ] && List.mem c.slug christmas_1_weekday_slugs
+  then Some "C30"
+  else if
+    (* C31 -- Task 9: the SAME rubric's own next sentence, "diebus
+       ferialibus a 7 ad 12 ianuarii ante dominicam I post Epiphaniam...
+       Missa de festo Epiphaniae". A SUBSET guard, not an exact match,
+       unlike C29/C30 -- this slug family's own civil dates ALREADY
+       carried a Season/Slug_f/Colour_f divergence before this task
+       (lectio's own season boundary starts Time-after-Epiphany at
+       Epiphany itself, RG 72-73's own later boundary colitur keeps;
+       lectio's own week numbering has no Holy Family to wait for, so its
+       own slug for these dates is already "ef-time-after-epiphany-1-
+       <weekday>", not "ef-christmas-2-..."), previously absorbed by C1's
+       own [jan_6_13_slug] allow-list (which already names this family).
+       First_f/Gospel_f now join an EXISTING divergence, they do not
+       create a new, cleaner one -- ONE of the two, not necessarily both:
+       a feria's Epistle and Gospel are independent lookups, and can each
+       happen to already coincide (e.g. the Gospel pericope shared with a
+       neighbouring office while the Epistle differs). [_with_saturday]
+       -- see that list's own comment for why Saturday belongs here for
+       THIS family (a step-3 ripple through the now-corrected
+       [ef-holy-name-sunday], not a direct entry of its own). *)
+    subset diffs [ Season; Slug_f; Colour_f; First_f; Gospel_f ]
+    && (List.mem First_f diffs || List.mem Gospel_f diffs)
+    && List.mem c.slug christmas_2_weekday_slugs_with_saturday
+  then Some "C31"
   else if is_bvm_saturday_row c diffs then Some "C17"
   else if diffs = [ Colour_f ] && List.mem c.slug audit_colour_corrected_slugs then Some "C18"
   else if
@@ -1054,6 +1165,45 @@ let layer_c_reason (l : row) (c : row) diffs =
     && subset diffs [ Colour_f; First_f; Gospel_f ]
   then Some "C19"
   else if
+    (* C32 -- Task 9: "Missa Dominicae Primae post Epiphaniam", Holy
+       Family's own week, the five non-Saturday weekdays. Checked AFTER
+       C19 deliberately: C19's own 7 "week shift" years already explain
+       this same slug family's own citation divergence there (a
+       DIFFERENT root cause, the week-numbering offset, not this task's
+       own fix) -- this entry must not re-claim those rows under a
+       citation-content reason C19 already owns, so its own population is
+       exactly the OTHER (non-shift-year) civil dates. A SUBSET guard,
+       the same reason as C31 immediately above: unlike C29/C30, THIS
+       slug family's own civil dates already carried a bare
+       Season/Colour_f divergence before this task (SAME slug on both
+       sides here, unlike C31 -- lectio's own season boundary alone,
+       RG 72-73 vs Epiphany), previously absorbed by C1's own generic
+       6-13 January window (no [Slug_f] involved, so C1's own
+       [jan_6_13_slug] gate was never even reached). First_f/Gospel_f now
+       join that existing divergence -- ONE of the two, the same reason
+       as C31's own note. [_with_saturday]: this family's own Saturday IS
+       reachable by a step-3 walk landing on [ef-holy-name-sunday]
+       (unlike [ef-christmas-1-saturday]'s own walk, which structurally
+       cannot reach that far forward -- confirmed, not assumed, by the
+       pinned test_lectionary_ef.ml case for that one slug). *)
+    subset diffs [ Season; Colour_f; First_f; Gospel_f ]
+    && (List.mem First_f diffs || List.mem Gospel_f diffs)
+    && List.mem c.slug time_after_epiphany_1_weekday_slugs_with_saturday
+  then Some "C32"
+  else if
+    (* C33/C34 -- Task 9: two major movable feasts of the Lord that had NO
+       lectionary entry at all before this task (tools/
+       bootstrap_lectionary.ml's own [movable_feast_entries]) -- lectio has
+       none either, so colitur and lectio used to independently fall back
+       to the SAME preceding-Sunday (Trinity's own) citation and silently
+       agree on the wrong answer, the "Holy Thursday violet in both" shape
+       CLAUDE.md's own validation-layers section names. Two ids, not one:
+       distinct feasts, distinct citations, the same discipline C33/C34's
+       neighbours already apply. *)
+    diffs = [ First_f; Gospel_f ] && String.equal c.slug "ef-corpus-christi"
+  then Some "C33"
+  else if diffs = [ First_f; Gospel_f ] && String.equal c.slug "ef-sacred-heart" then Some "C34"
+  else if
     diffs = [ First_f; Gospel_f ] && String.equal l.first "-" && String.equal l.gospel "-"
     && List.mem c.slug lectio_no_reading_slugs
   then Some "C20"