tests/test_gitapex_gate_local_preflight.py: stale "28 exclusions" prose count (registry now has 29) #1228

Facts

• Found while running the full repository test suite on main's own
  current tip (3da9cb2, isolated in a separate worktree) as part of
  verifying a merge of main into PR fix(hooks): fail closed on missing/malformed jq in four PreToolUse gates #1213 (issue hooks/: four more hooks fail open on missing/malformed jq (same class as gitapex#436) #1208) didn't
  introduce any regression.
• tests/test_gitapex_gate_local_preflight.py::test_no_prose_count_contradicts_the_registry
  fails on main's own tip, independent of PR fix(hooks): fail closed on missing/malformed jq in four PreToolUse gates #1213 or anything in this
  session -- confirmed by running it against an isolated worktree
  checked out directly at origin/main, with no local changes applied:
  AssertionError: prose counts contradicting .gitapex/ssot.json: tests/test_gitapex_gate_local_preflight.py: says 28 exclusions, registry has 29.
• The stale prose is at tests/test_gitapex_gate_local_preflight.py:880,
  inside test_every_unwired_gate_records_why's own docstring: "a gate
  with no working-tree form must say so in prose, so the 28 exclusions
  stay readable as decisions rather than as coverage silently lost."
  .gitapex/ssot.json currently has 29 gates with no "local" in their
  planes, not 28 -- two recently-merged PRs each registered one new
  non-local gate (merge-pull-request-block via PR refactor(hooks): consolidate hooks.json description into ssot.json registry #1223/issue refactor(hooks): consolidate hooks.json description into ssot.json gate registry #1222,
  and pr-duplicate-issue via the PR closing issue fix(dedup): add deterministic duplicate-work checks for PRs, retrospective issues, and new issues #1197), and at least
  one of them didn't update this specific hardcoded prose count even
  though test_no_prose_count_contradicts_the_registry exists
  specifically to catch this class of drift.
• Not investigated: whether main's CI is currently red on this check,
  or whether it's excluded from the default run somehow. This issue
  reports the fact and its precise location; it does not itself audit
  CI's current pass/fail state.

Requested outcome

Update the digit at tests/test_gitapex_gate_local_preflight.py:880
from 28 to 29 (or rephrase to reference the count structurally
instead of a hardcoded digit, if that's a better long-term fix -- left
as an implementer's judgment call, not mandated here) so
test_no_prose_count_contradicts_the_registry passes again on main.

Acceptance Criteria Map

Constraints

• Fix only the stale count; do not change test_no_prose_count_contradicts_the_registry's own detection logic.

Non-goals

• Does not audit whether other prose counts elsewhere in the repository
  have drifted the same way -- this issue is scoped to the one instance
  this session's own full-suite run actually caught.

Note

Root-caused directly against main's own tip in an isolated worktree,
not inferred from PR #1213's own (unrelated) diff -- this is genuinely
pre-existing on main, confirmed before filing.