test(ef): reach RG 91 entry 27 through the real pipeline, end-to-end

Entry 27's band branch has been unreachable for the right reason since
it was written: precedence_ef.ml's own comment on it acknowledged the
row existed, but the one test exercising it built a hand-typed
candidate, never a real Temporal_ef.temporal output, and nothing built
the office itself. Closes that gap on both sides:

- The band-table row now sources its candidate from Temporal_ef.temporal
  directly (of_temporal), the same review-finding-3 discipline the
  Ember/vigil rows already use, so a future regression in the office's
  own construction fails this row too, not only the unit-level test.
- Two new resolve-level end-to-end tests, using the real band/
  disposition/admit triple (not stubs): the office winning a real
  otherwise-unoccupied Saturday and admitting an ordinary commemoration
  alongside it (1 August 2026, Holy Machabees -- the task brief's own
  worked oracle example), and the office genuinely losing to a real
  competing feast and being omitted under RG 26, with no confusion from
  the one live data witness that also carries subject Bvm (12 September,
  Most Holy Name of Mary) -- proving RG 26's rank-keyed omission fires
  before any subject-keyed rule ever could.
- Two golden pins mirroring the same two dates, checked by hand against
  `colitur day 2026` before being typed in, following this file's own
  "a golden test blesses whatever it is given, verified independently"
  discipline. The existing 2038-03-06 pin (RG 91 entry 24 outranking a
  Class4 candidate) is corrected in passing: that date is itself a
  Saturday, so the displaced candidate the pin's own omitted-list check
  names is now precisely entry 27's own shape, not entry 28's bare
  catch-all -- the win/lose outcome the pin proves is unchanged, only
  its own more precise citation.

Mutation-tested (recorded in the task report): reverting
is_bvm_saturday to a constant false reddens the dedicated unit test, the
new end-to-end resolve test, and the 2026-08-01 golden pin, while
leaving the 2026-09-12 "loses to a real feast" pin untouched, exactly as
expected since that date's outcome never depended on the fix.

1 files changed, 81 insertions, 2 deletions

diff --git a/test/test_golden.ml b/test/test_golden.ml
index 594ea6b..bb4bcfc 100644
--- a/test/test_golden.ml
+++ b/test/test_golden.ml
@@ -851,8 +851,22 @@ let test_iii_class_feast_wins_class4_feria () =
     "2008-04-05 saturday season=paschaltide week=2 slug=vincent-ferrer rank=class-3 colour=white subject=saint name_la=- comms=[] in=- \
      out=[]";
   let d4 = fetch 2038 3 6 in
+  (* ef-bvm-saturday task: 6 March 2038 is a SATURDAY, so the displaced
+     temporal candidate is now RG 78's own BVM Saturday office
+     ({!Rite_ef.Temporal_ef}'s own [bvm_saturday_names] citation) rather than
+     a bare ferial -- but per that citation's own "Slug" paragraph the SLUG
+     is deliberately left unchanged (reused, not replaced), precisely so
+     this test's own [omitted_has] probe still needs no update: the displaced
+     candidate is still named "ef-septuagesima-2-saturday", only its
+     colour/subject/name would have differed had it won. RG 91 entry 24
+     (Sts Perpetua and Felicity) still outranks it either way -- entry 24 <
+     entry 27, both above entry 28 -- so the disposition/win outcome this
+     test proves is genuinely unchanged by this task, now more precisely
+     entry 24 > entry 27 rather than > 28 for this one date (a Saturday), not
+     that the test itself needed touching. *)
   Alcotest.(check bool)
-    "2038-03-06: the displaced class-4 Septuagesima feria is in [omitted] (RG26), proving RG91 entry24>28 fired"
+    "2038-03-06: the displaced class-4 feria (RG78's own BVM-Saturday office, \
+     for this Saturday) is in [omitted] (RG26), proving RG91 entry24>27 fired"
     true
     (omitted_has d4 "ef-septuagesima-2-saturday");
   Alcotest.(check string)
@@ -869,6 +883,63 @@ let test_iii_class_feast_wins_class4_feria () =
     "2038-03-09 tuesday season=septuagesima week=3 slug=frances-rome rank=class-3 colour=white subject=saint name_la=- comms=[] in=- \
      out=[]"
 
+(* ------------------------------------------------------------------ *)
+(* RG 91 entry 27 / RG 78-79 (Caput IX, "De sancta Maria in sabbato") --
+   ef-bvm-saturday task, Rite_ef.Temporal_ef's own [bvm_saturday_names]
+   citation has the full primary-source argument. Two real-data pins, the
+   task brief's own worked example and its mirror-image negative case,
+   matching this file's own header rule: EVERY literal below was checked by
+   hand against `colitur day 2026` before being typed in here (this task's
+   own report has the full session log), never copied from a run and
+   rationalised after. *)
+
+(* The task brief's own worked example, checked against the oracle
+   (missalemeum) directly: "2026-08-01|4|w|V Mass of the B. V. M. -- Salve,
+   Sancta Parens|Saturday_after_IX_Sunday_after_Pentecost|Holy Machabees".
+   colitur's own rank (4) and colour (white) now match; the oracle's own
+   numbered-Mass title (RG 309(a)'s own "iuxta temporum diversitatem", not
+   modelled here -- {!bvm_saturday_names}'s own "Slug" paragraph) and the
+   observed day's own English identity remain outside what this comparator
+   can see (M18's own limit) -- this pin instead proves what colitur DOES
+   compute: white/class-4/subject Bvm/the Latin title, and Holy Machabees
+   (real data/ef/sanctoral.sexp entry, Commemoration_only, Class3, 1 August)
+   still admitted as an ORDINARY commemoration alongside it, matching the
+   oracle's own commemoration list exactly. *)
+let test_bvm_saturday_2026 () =
+  check ~msg:"2026-08-01: the BVM Saturday office observed, white overriding Time after Pentecost's green, \
+              Holy Machabees commemorated -- matches the task brief's own oracle example"
+    2026 8 1
+    "2026-08-01 saturday season=time-after-pentecost week=9 slug=ef-time-after-pentecost-9-saturday \
+     rank=class-4 colour=white subject=bvm name_la=Officium sanctae Mariae in sabbato \
+     comms=[holy-machabees:ordinary] in=- out=[]"
+
+(* The mirror-image case: RG 78's own protasis is FALSE here (a real feast
+   wins the Saturday outright), so the BVM-Saturday-shaped temporal
+   candidate must lose and be omitted under RG 26, exactly like the plain
+   ferial candidate it replaces always was -- no commemoration at all,
+   proving [band]/[disposition] need no special case for this office (the
+   design claim the task report makes explicitly). ALSO the one live data
+   witness where the winning candidate itself carries [subject = Bvm]
+   (data/ef/adjustments.sexp's own `Edit most-holy-name-of-mary
+   ((Set_subject Bvm))`) -- proving no Bvm-vs-Bvm confusion with a
+   RG112(a)-style subject rule: [omitted_has] below confirms the loser is
+   genuinely reported omitted (RG26 fires on [rank = Class4] alone, checked
+   ahead of any subject-based branch in {!Rite_ef.Precedence_ef.disposition}),
+   not silently swallowed by a same-subject exclusion that does not exist
+   for [Bvm] anywhere in this codebase. *)
+let test_bvm_saturday_loses_to_most_holy_name_of_mary_2026 () =
+  let d = fetch 2026 9 12 in
+  Alcotest.(check bool)
+    "2026-09-12: the displaced BVM-Saturday-shaped candidate is in [omitted] (RG26), not silently absent"
+    true
+    (omitted_has d "ef-time-after-pentecost-15-saturday");
+  Alcotest.(check string)
+    "2026-09-12: Most Holy Name of Mary (III class, subject Bvm, data/ef/adjustments.sexp) outranks the \
+     class-4 BVM-Saturday candidate; no commemoration at all"
+    "2026-09-12 saturday season=time-after-pentecost week=15 slug=most-holy-name-of-mary rank=class-3 \
+     colour=white subject=bvm name_la=- comms=[] in=- out=[]"
+    (describe d)
+
 let suite =
   ( "golden pins (known-tricky years)",
     [ Alcotest.test_case "Easter extreme: 1598 earliest (22 Mar, Gauss-verified)" `Quick
@@ -925,5 +996,13 @@ let suite =
       Alcotest.test_case
         "ef-rebootstrap: RG91 entry24 (III-class feast) outranks entry28 (class-4 feria) -- the six dates \
          deferred since 2038's own Easter-extreme pin (2008-04-02/04/05, 2038-03-06/08/09)"
-        `Quick test_iii_class_feast_wins_class4_feria
+        `Quick test_iii_class_feast_wins_class4_feria;
+      Alcotest.test_case
+        "RG91 e27/RG78: the BVM Saturday office wins, Holy Machabees commemorated -- matches the task \
+         brief's own oracle example (2026-08-01)"
+        `Quick test_bvm_saturday_2026;
+      Alcotest.test_case
+        "RG91 e27/RG78/RG26: the BVM Saturday office loses to Most Holy Name of Mary (subject Bvm too) \
+         and is omitted, no confusion (2026-09-12)"
+        `Quick test_bvm_saturday_loses_to_most_holy_name_of_mary_2026
     ] )