C-i needs a positional qualifier, and the never-given particles get evidence records

[derek73]

Two commits. Comments and one criterion correction — membership is
byte-identical, 28 never-given members before and after, 3459 tests
green.

C-i was wrong on de

C-i as recorded yesterday asked whether a word is borne as an ordinary
name in some tradition. Applied to this set's most load-bearing member
it self-destructs:

parse("Bimal De").family == "De"       ← borne Bengali/Odia surname

So C-i said mark de ambiguous, which would break de la Vega and
every leading-particle reading.

The missing qualifier is position, and the existing set was already
obeying it. P1 acts on the leading position or a lone particle in the
given role, so what matters is whether the naming use occupies that
position:

van   Vietnamese Văn in given position    collides    ambiguous  ✓
bar   Bar Refaeli, given position         collides    ambiguous  ✓
do    Đỗ leads a surname                  collides    ambiguous  ✓
de    "De" is a TRAILING surname          no clash    never-given ✓

It reverses a verdict, and the old reading would have declined a fix

das was called ambiguous on #360 under the old reading. Measured:

Anjali Das        unchanged by never-given `das`
Bimal Das         unchanged
Maria das Neves   family='Neves' today  →  family='das Neves'   ← a fix

The surname is trailing, the rule acts leading. #360's table is
corrected in a comment there, since it was public.

The records

The never-given side is the falsifiable claim — one attested bearer in
the acting position overturns it — so it is the side worth recording.
"Ambiguous" is the safe default and needs no argument.

Each entry now states what the word is (article, preposition,
patronymic marker), because that is the only form a never-given record
can take: one bearer settles the ambiguous side, no amount of searching
settles absence.

A bare entry means nobody has examined it. Absence of a comment is
the audit record, so #360's remaining work is visible at a glance
without a separate ledger. Style follows the #269 Arabic/Hebrew blocks,
which already did this for their own members.

Deliberately not included: the ambiguous members #360 lists as
candidates to move (mc, ste, aan, aen, heer, te, tho,
thoe, vel, vande). Their records are their judgments, and #360
hasn't made them — a comment now would pre-judge the issue. They stay
bare.

How the flaw was caught

Writing a per-word record for de is what falsified the criterion. A
verdict alone hides its reasoning; a record has to state why, which
forces the collision question. C-i had survived three review agents and
a merge because nothing tested it against de — and the error pointed
at the safe side, so it could never have surfaced as a misparse.

That is the argument for the records existing, one word into writing
them.

Verification

uv run --frozen pytest — 3459 passed, 20 skipped, 11 xfailed. ruff
clean. len(particles - particles_ambiguous) == 28 before and after.

Note these are plain # comments, not #: doc-comments, so they are
internal and do not reach the rendered API reference.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.51%. Comparing base (7e8c8f5) to head (558ff4c).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #392   +/-   ##
=======================================
  Coverage   98.51%   98.51%           
=======================================
  Files          44       44           
  Lines        2895     2895           
=======================================
  Hits         2852     2852           
  Misses         43       43           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.