kernel(rite-ef): apply RG16(a) -- a Feast of the Lord leaves the impeded Sunday uncommemorated

Caput III "De Dominicis", RG 16(a): a Feast of the Lord, I or II class,
occurring on a II-class Sunday takes the Sunday's own place with all its
rights and privileges -- of the Sunday, therefore, no commemoration is
made. band's entry 14 already ranked such a feast above the Sunday
correctly; disposition had no exception for the collision, so the impeded
Sunday fell through to the ordinary RG 95/109(a) commemorate-or-omit
catch-all and was wrongly admitted as a privileged commemoration.

Adds a branch to Rite_ef.Precedence_ef.disposition, gated on all four of:
loser is temporal-origin, loser rank Class2, loser carries the Sunday slug
marker, winner rank Class1-or-2 with subject Lord. No signature change --
disposition already receives the winner (RG 33's own vigil-omission branch
already reads it); each conjunct is proved independently load-bearing by
a dedicated disposition_cases row, including one real-data case each for
the winner's subject and rank conjuncts, and a defensive synthetic case
for the winner rank conjunct's currently-unreachable Class3 shape. A new
end-to-end resolve-level test (2028-08-06, the Transfiguration) proves the
real band/disposition pipeline together, checking the Omit reason string
distinguishes this from admit's own admission-limit cut.

1 files changed, 72 insertions, 7 deletions

diff --git a/lib/rites/rite_ef/precedence_ef.ml b/lib/rites/rite_ef/precedence_ef.ml
index db7e708..179fa13 100644
--- a/lib/rites/rite_ef/precedence_ef.ml
+++ b/lib/rites/rite_ef/precedence_ef.ml
@@ -244,13 +244,14 @@ let band (ctx : Vocab_ef.season Precedence.context) (c : Vocab_ef.rank Precedenc
 
 (* Task 8: what happens to the day's LOSING candidate (docs/research/
    rules-register.md §4, "Occurrence" RG 92-95 and "Vigils" RG 33, plus RG
-   94). [band] above decides who wins; this decides the loser's fate, which
-   turns on the LOSER's own rank and status (RG 95), except RG 33's vigil
-   omission, which also has to read the winner. Nothing here ever returns
-   [Precedence.Repose]: that disposition denotes RG 100-102's *repositio*
-   (perpetual impediment from a proper/diocesan calendar), out of this
-   plan's scope -- see calendar.mli's own note that nothing in the EF
-   ruleset currently emits it. *)
+   94; also §6.0/Caput III "De Dominicis" RG 16(a), below). [band] above
+   decides who wins; this decides the loser's fate, which turns on the
+   LOSER's own rank and status (RG 95), except RG 33's vigil omission and RG
+   16(a)'s Sunday-suppression, which also have to read the winner. Nothing
+   here ever returns [Precedence.Repose]: that disposition denotes RG
+   100-102's *repositio* (perpetual impediment from a proper/diocesan
+   calendar), out of this plan's scope -- see calendar.mli's own note that
+   nothing in the EF ruleset currently emits it. *)
 
 (* RG 33 -- CORRECTED 2026-08-12 (Task 16, primary-source-verified against
    docs/research/1962-06-23,_SS_Ioannes_XXIII,_Missale_Romanum,_LT.pdf, the
@@ -605,6 +606,70 @@ let disposition ~(winner : Vocab_ef.rank Precedence.candidate)
        register's own §4 "Vigils" entry states RG 32 only as "same pattern
        [as RG 31]", not verbatim -- now closed here). *)
     Precedence.Omit
+  else if
+    is_temporal
+    && cel.Celebration.rank = Class2
+    && is_sunday_slug (Slug.to_string cel.Celebration.slug)
+    && (let wcel = winner.Precedence.cel in
+        (wcel.Celebration.rank = Class1 || wcel.Celebration.rank = Class2)
+        && wcel.Celebration.subject = Subject.Lord)
+  then
+    (* RG 16(a) (docs/research/rules-register.md §6.0, Caput III "De
+       Dominicis", primary text): "Dominica II classis, in occurrentia,
+       festis II classis praefertur. Attamen: a) festum Domini I aut II
+       classis, in dominica II classis occurrens, locum tenet ipsius
+       dominicae cum omnibus iuribus et privilegiis: de dominica, proinde,
+       NULLA FIT COMMEMORATIO" -- a Feast of the Lord, I or II class,
+       occurring on a II-class Sunday, takes the Sunday's own place with all
+       its rights and privileges: OF THE SUNDAY, THEREFORE, NO
+       COMMEMORATION IS MADE. {!band} entry 14 already ranks such a feast
+       above the Sunday (RG 91 entry 14 < entry 15), so the feast is
+       correctly [observed]; this is [disposition]'s own answer for what
+       becomes of the SUNDAY once it has lost -- [Omit], not the [Commemorate]
+       every other impeded II-class Sunday gets via RG 109(a)/RG 111(b)
+       below. This is the ONE place in this function that needs the WINNER's
+       [subject], not only the loser's own fields -- like RG 33's vigil
+       omission above, not a new kind of signature: [Precedence.rules.disposition]
+       already takes [~winner], this is simply its first other reader.
+
+       Both winner-side conjuncts are load-bearing, proved by
+       [test_precedence_ef.ml]'s own paired rows (the brief's "one without
+       the other proves nothing"):
+       - [rank = Class1 || rank = Class2]: {!band} entries 11-13 admit ANY
+         I-class feast -- Lord or Saint alike, no subject test at all (entry
+         14's own comment: unlike entry 16, entries 11-13 draw no such
+         line) -- ahead of a II-class Sunday. REACHABLE at Class1 on real
+         data: e.g. the Nativity of St John the Baptist (24 June, I class,
+         Saint) landing on a Time-after-Pentecost Sunday already wins the
+         day under {!band} alone; without this conjunct RG 16(a) would
+         wrongly fire for it too. (At Class2 this conjunct adds no further
+         cases beyond what the [subject = Lord] conjunct below already
+         requires -- entry 16's ordinary II-class feasts never outrank a
+         Sunday in {!band} in the first place, 16 > 15 -- so it is only
+         independently reachable at Class1.)
+       - [subject = Lord]: at Class2 this is what {!band} entry 14 already
+         requires of its own winners, so it adds no further restriction
+         there; at Class1 it is independently reachable and necessary, per
+         the St John Baptist example above -- dropping it would fire RG
+         16(a) for any winning I-class feast at all, exactly the
+         over-wide branch the task brief warns against.
+
+       The SUNDAY-side conjuncts: [rank = Class2] excludes a I-class Sunday
+       (Advent/Lent/Passiontide, Low Sunday, {!band} entry 6) -- RG 16(a)'s
+       own text says "Dominica II classis", not "any Sunday", and a I-class
+       Sunday can never actually reach here as a [loser] against a
+       Class1-or-2 Lord winner in the first place: entry 6's own band value
+       (6) is lower than every entry a Lord feast of Class1 or Class2 could
+       occupy (3, 11-14), so a I-class Sunday always wins outright -- this
+       is exactly why the rubric restricts itself to II class. [is_sunday_slug]
+       (the same marker RG 33's [impedes_vigil] and RG 109(a)'s
+       [privilege_of] already use, with no [is_temporal] guard there either
+       -- no sanctoral slug this codebase's data produces contains
+       {!sunday_marker}) is the "Dominica" half; [is_temporal] is kept
+       alongside it anyway, the same explicit-but-not-load-bearing defence
+       the RG26 branch above gives its own [not is_vigil] guard, rather than
+       relying on that absence silently. *)
+    Precedence.Omit
   else
     (* RG 95's other branch: "aut commemorantur aut penitus omittuntur" --
        commemorated or wholly omitted. Reached by every SANCTORAL loser