fix(of): correct subject over-match, provenance truth, and stale hazard comment

Four fixes from fix round 2 review:

classify_subject's "maria"/"b.m.v" substring test over-matched 12 saints
who merely carry "Maria" in their own name (Maximilian Mary Kolbe, John
Mary Vianney, Mary Magdalene among them) as Bvm. Replaced with an
explicit BVM_DATES table of the 13 genuinely Marian entries. Bvm 25 -> 13.

The provenance header claimed lectio is "never a source", which omitted
that slugs and English names both come from it. Corrected to state
exactly what does (slugs, en names) and does not (every substantive
field) come from lectio. A missing lectio file used to degrade silently
to a different SHA-256 with no en names and no warning; it now fails
loudly. lectio's own SHA-256 is now pinned alongside the Missal PDF's.

Four more entries carried a second pdftotext artifact distinct from the
letter-spacing already handled -- a combining diacritic rendered as a
stray spacing character (Kim Taegŏn/Chŏng Ha-sang, Đũng Lạc, Makhlūf,
Kęty). Hand-repaired the same way Peter Damian was, each corroborated by
the identical corruption recurring in the Missal's own index.

precedence_of.ml's PHASE 3 HAZARD comment about All Souls (2 November)
was asserting a gap that no longer exists now that OF sanctoral data
ships. Added test_all_souls_beats_an_ordinary_sunday, which resolves the
shipped All Souls entry against a real per-annum Sunday (2025-11-02)
through the actual Precedence.resolve pipeline, and updated the comment
to record the resolution instead of demanding it.

1 files changed, 38 insertions, 19 deletions

diff --git a/lib/rites/rite_of/precedence_of.ml b/lib/rites/rite_of/precedence_of.ml
index ab5e394..0587f0a 100644
--- a/lib/rites/rite_of/precedence_of.ml
+++ b/lib/rites/rite_of/precedence_of.ml
@@ -140,25 +140,44 @@ let band (ctx : Vocab_of.season Precedence.context)
     30 (* Tabula I.3 -- universal or temporal-origin, never proper; see the
           [is_proper] comment above for why this is not [is_universal].
 
-          ENTRY 3 HAS A SECOND CLAUSE THIS BRANCH DOES NOT IMPLEMENT (transcription
-          line 840, identical in the 1969 Calendarium at OCR line 936):
-          "Commemoratio omnium fidelium defunctorum" -- All Souls, 2 November.
-          Nothing constructs it: there is no sanctoral data yet (Phase 3), and
-          Vocab_of.rank has no constructor for it -- the 2002 Calendarium prints
-          2 November UNGRADED ("Omnium Fidelium defunctorum" / "Commemoratio
-          omnium Fidelium defunctorum"), so the natural rank tag is ambiguous:
-          Memoria_obligatoria (band 100, loses to a per annum Sunday, WRONG),
-          Memoria_ad_libitum (band 120, worse), or Sollemnitas (band 30, the
-          right precedence but a rank word the source never applies to this
-          day). The 1969 Calendarium's own printing of this clause additionally
-          carries a qualifier the 2002 text drops -- "quae tamen locum cedit
-          dominicae occurrenti" -- an edition difference Phase 3 will also need.
-          PHASE 3 HAZARD: transcribing 2 November with the natural ungraded
-          rank will silently make it an ordinary weekday whenever it falls on a
-          Sunday, with every test in this branch still green, because no test
-          here knows entry 3 has a second clause. Phase 3 must settle how this
-          day is tagged (or add a dedicated slug test here, mirroring
-          [entry_2_named]) before shipping OF sanctoral data. *)
+          ENTRY 3'S SECOND CLAUSE -- RESOLVED, fix round 2, Task 1
+          (2026-08-25; transcription line 840, identical in the 1969
+          Calendarium at OCR line 936): "Commemoratio omnium fidelium
+          defunctorum" -- All Souls, 2 November. The 2002 Calendarium
+          prints 2 November UNGRADED ("Commemoratio omnium Fidelium
+          defunctorum"), so no grade word could be transcribed literally;
+          data/of/calendar-2002.sexp (tools/extract_of_calendar.py's own
+          ALL_SOULS_OVERRIDE) tags it [Sollemnitas] anyway, on this table's
+          own authority -- entry 3 itself names "Commemoratio omnium
+          fidelium defunctorum" in the SAME line as "Sollemnitates ...
+          Sanctorum in Calendario generali inscriptae", i.e. this table
+          places it at THIS precedence tier regardless of what word the
+          calendar page prints next to it. (For the record, the two
+          rejected alternatives were Memoria_obligatoria, band 100, loses
+          to a per annum Sunday, WRONG; and Memoria_ad_libitum, band 120,
+          worse.) Proved, not merely asserted:
+          test/test_precedence_of.ml's [test_all_souls_beats_an_ordinary_
+          sunday] loads the SHIPPED "all-souls" entry and resolves it
+          against a real per-annum Sunday (2 November 2025, itself a
+          Sunday) through {!Colitur_kernel.Precedence.resolve} and
+          [rules] below -- not [band] in isolation -- confirming All Souls
+          is [observed], not the Sunday.
+
+          STILL OPEN, deliberately out of this task's scope: the 1969
+          Calendarium's own printing of this clause carries a qualifier
+          the 2002 text drops -- "quae tamen locum cedit dominicae
+          occurrenti" -- an edition difference not reconciled here.
+
+          A LOSING All Souls -- unreachable on universal data (nothing
+          shipped outranks a Tabula I.3 Sollemnitas except the Triduum's
+          own Easter-bound window, which can never reach 2 November) but
+          real the moment an overlay puts something higher there --
+          would be TRANSFERRED, not omitted: [disposition] below sends
+          any losing Sollemnitas forward via Normae n. 60, and nothing
+          distinguishes "Commemoratio" from an ordinary Sollemnitas for
+          that purpose. The same shape as [Rite_ef.Precedence_ef]'s own
+          [entry_14_movable_band] (CLAUDE.md, "Carried into Plan 4") --
+          recorded here rather than built defensively against. *)
   else if c.cel.rank = Vocab_of.Sollemnitas && is_proper layer then
     40 (* Tabula I.4 *)
   else if c.cel.rank = Vocab_of.Festum && c.cel.subject = Subject.Lord