fix(citation): close the final review's blocking findings

The branch was RED and reported green. `dune test` exited 1: test/cli.t
pinned the pre-fix output `kings_1 19:3-8`, which the previous commit had
already fixed to `1 Reg 19:3-8`. The gate command piped dune through
`tail`, so it reported tail's exit status, and cram prints its diff
BEFORE the alcotest summary, so the two lines shown were the passing
ones. Verify with `dune test; echo $?`, never through a pipe.

A style file's own `book` key was unreachable. sigla_book resolved
against a hardcoded "abbr" and the result was applied unconditionally,
so the documented `[sigla] book = full` could never win. Render gains
book_string, and the style's own value is now the default that a flag or
config overrides. The unit test pinned style_of_fields correctly while
the wiring defeated it.

`lang --check` filtered the reference set to the celebration prefix, so
a file with no [bible] section at all reported a clean bill of health --
contradicting both the reason the keys change was made and lang.ml's own
comment. It now reports missing book names too.

The token test missed a FOURTH citation-bearing file: adjustments.sexp
writes citations as `Set_citation`, not `(reference ...)`. Its 16
citations all parse, so nothing was broken, but nothing was checking.
The first attempt at this fix read the file and extracted NOTHING -- the
marker stopped before the opening quote, so every payload was the part
label -- which is recorded in the code rather than left as a trap.

Also: colitur-config(5) claimed a trailing period the data does not
carry, and two la.ini scan quotes silently corrected OCR damage
("Ionae 3, I - I O", "Epistolse") while presenting themselves as
verbatim. Both are now marked as corrections.

2 files changed, 41 insertions, 14 deletions

diff --git a/test/cli.t b/test/cli.t
index d9d8ec5..b44a256 100644
--- a/test/cli.t
+++ b/test/cli.t
@@ -1261,12 +1261,14 @@ slug the engine can produce over 2020-2045:
 
   $ colitur lang --check d.ini
   d.ini: 725 of 725 celebrations named, 0 missing, 0 unknown
+  d.ini: 104 of 104 book names, 0 missing
 
 `--check` reports what is MISSING (a real slug with no entry):
 
   $ printf '[meta]\nlang = zz\n[celebration]\nef-epiphany = Test\n' > partial.ini
-  $ colitur lang --check partial.ini | tail -1
+  $ colitur lang --check partial.ini | tail -2
   partial.ini: 1 of 725 celebrations named, 724 missing, 0 unknown
+  partial.ini: 0 of 104 book names, 104 missing
 
 `--check` REJECTS an unknown slug (exit 1), so a typo is visible rather than
 silently dead -- its author would otherwise never learn why the name they
@@ -1431,15 +1433,20 @@ Latin-convention punctuation. The synthetic file below overrides only
 
 `--sigla-tradition` renumbers which book an id DENOTES (lang/traditions.ini),
 independently of style or naming -- `modern` maps `3 Kings` onto the id
-`kings_1`. Task 10's `la.ini` now carries a `[bible]` row for `kings_1` (a
-tradition target the Vulgate data never cites directly, book.mli), but only
-its own bare id -- an honest UNSOURCED placeholder, not a name (la.ini's own
-`[bible]` header note) -- so it still prints literally, now BY DESIGN rather
-than by the row's absence, still a real, if plain, witness that the
-tradition actually applied rather than a no-op:
+`kings_1`, whose own Latin name la.ini marks CONSTRUCTED: the 1962 Missal
+uses Vulgate numbering throughout, so it can contain no incipit for a book
+that exists only under a later convention. In Latin the modern tradition
+therefore only really moves Kings and Esdras -- Osee, Ionas, Ecclesiasticus
+and the Apocalypse keep their Vulgate names either way, because modern
+numbering is a vernacular convention:
 
   $ colitur readings 2027 --sigla-tradition modern | grep '^2027-02-17'
-  2027-02-17 ef-lent-ember-wed | kings_1 19:3-8 | Matth 12:38-50 | Feria IV Quatuor Temporum Quadragesimae
+  2027-02-17 ef-lent-ember-wed | 1 Reg 19:3-8 | Matth 12:38-50 | Feria IV Quatuor Temporum Quadragesimae
+
+Under an English file the same mapping shows its usual face:
+
+  $ colitur readings 2027 --lang en --sigla-tradition modern | grep '^2027-02-17'
+  2027-02-17 ef-lent-ember-wed | 1 Kgs 19:3-8 | Matt 12:38-50 | Lenten Ember Wednesday
 
 `table`/`render`, `emit` and `publish` accept the same three flags too --
 smoke-tested for exit status alone here (a minimal inline template, the
diff --git a/test/test_citation.ml b/test/test_citation.ml
index 1efaa10..133553b 100644
--- a/test/test_citation.ml
+++ b/test/test_citation.ml
@@ -117,8 +117,7 @@ let starts_with_at content pos prefix =
    [Str]/regex -- a plain forward scan for the marker, then read to the
    closing quote. Mirrors the coordinator's own survey command (grep -oh
    over the reference marker, quote-delimited). *)
-let references content =
-  let marker = "(reference \"" in
+let references_with marker content =
   let mlen = String.length marker in
   let len = String.length content in
   let rec loop pos acc =
@@ -151,14 +150,35 @@ let book_token r =
   let stop = if !i < len && r.[!i] = '.' then !i + 1 else !i in
   String.sub r 0 stop
 
+(* A citation is written TWO ways in this project's data, and a survey that
+   knows only one of them silently under-reads the corpus:
+     (reference "Isa 60:1-6")            -- lectionary/sanctoral/commons
+     (Set_citation First "Wis 7:7-14")   -- adjustments.sexp, an overlay
+   Both markers are scanned below. *)
+let references content =
+  (* Each marker must include the OPENING QUOTE. [references_with] returns
+     the text between the marker and the next '"', so a marker stopping at
+     "(Set_citation " yields "First " / "Gospel " -- the part label, never
+     the citation -- and every entry is then silently discarded. That was
+     this function's first version, and it read the file while extracting
+     nothing at all. *)
+  references_with "(reference \"" content
+  @ references_with "(Set_citation First \"" content
+  @ references_with "(Set_citation Gospel \"" content
+
 let test_every_data_file_token_resolves () =
   (* The check whose absence caused fix round 1: the brief surveyed only
      the lectionary and missed 21 tokens, several common, living in
-     sanctoral.sexp and commons.sexp. Read all three files at test time and
-     re-derive the token set from them, rather than hardcoding a list, so
-     this keeps working when the data changes. *)
+     sanctoral.sexp and commons.sexp.
+
+     adjustments.sexp was then missed AGAIN, by this very test, because it
+     writes citations as `Set_citation` rather than `(reference ...)` -- the
+     same "one file too few" shape twice over. All FOUR shipped files are
+     read here, and the token set is re-derived from them at test time
+     rather than hardcoded, so this keeps working when the data changes. *)
   let files =
-    [ "../data/ef/lectionary.sexp"; "../data/ef/sanctoral.sexp"; "../data/ef/commons.sexp" ]
+    [ "../data/ef/lectionary.sexp"; "../data/ef/sanctoral.sexp";
+      "../data/ef/commons.sexp"; "../data/ef/adjustments.sexp" ]
   in
   let tokens =
     files