docs(apps/docs): fix stale docblock in page.tsx breadcrumb walk

[baozhoutao]

Fixes #13949

What

apps/docs/app/[lang]/docs/[[...slug]]/page.tsx's docsTrail() docblock carried two
errors, both fixed in this one edit, one file, prose only:

1. Stale census. The paragraph claimed "8 short trails remain … its meta.json
   still lists "index". The condition is therefore live, just rare." That went false
   when PR docs(nav): stop listing "index" in content/docs/releases/meta.json #13946 removed "index" from content/docs/releases/meta.json — today,
   zero meta.json under content/docs lists "index" in pages
   (git grep -c '"index"' HEAD -- 'content/docs/**/meta.json' → no hits). Per the
   triage's measured lesson (a count that rotted within a day), the paragraph is
   rewritten to the mechanism only: how a URL-less ancestor arises, that docs site: 172 of 403 doc pages get a breadcrumb that skips its section — meta.json listing "index" detaches the folder index #12352
   and docs(nav): stop listing "index" in content/docs/releases/meta.json #13946 fixed every known instance, and that the condition is structural and
   can recur — no new count written in its place.
2. Misattribution, independent of docs(nav): stop listing "index" in content/docs/releases/meta.json #13946 and wrong on origin/main today:
   the docblock said the walk "drops" the un-linkable ancestor. getBreadcrumbItems()
   does not drop it — it emits the ancestor with url: undefined. The drop is local:
   docsTrail()'s own loop guard, if (typeof item.name !== 'string' || !item.url) continue;, is what skips it.

No behavior change: docsTrail()'s logic, the breadcrumb loop, and the "the missing
URL is deliberately not reconstructed here" rule paragraph (unaffected per the triage)
are untouched. git diff --stat on this branch touches exactly one file.

Why this needed its own PR

Per the issue: content/docs/releases/** is fenced unconditionally in PR #13946's
dispatch, whose legality rested on "one file, nothing else" — apps/docs/** would
have been a second file. The fence that made #13946 possible is what prevented it
from carrying this correction; this PR is the follow-on the triage said was needed.

Tests

• pnpm --filter @objectstack/docs typecheck (fumadocs-mdx && next typegen && tsc --noEmit) — PASS, both before opening (on the pre-merge tree) and again after
  merging origin/main into this branch.
• node scripts/check-docs-nav-label.mjs — PASS (9-clause battery, the gate this
  docblock's neighboring includePage flip is pinned by).
• node scripts/check-docs-nav-label.mjs --self-test — PASS (25 assertions).
• pnpm exec eslint --no-inline-config --format json narrowed to the one edited file
  — PASS, 0 errors/warnings (repo ESLint has no type-aware linting anywhere, so this
  narrowing moves no untouched file's judgment).
• The 9 gate families node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack --commands derived for this diff — all PASS (check:docs-locale-catch-all,
  check:logger-receiver-detach, check:page-declaration-shape, check:published-files,
  check:test-source-alias, check:type-source-resolution, plus the two
  check-docs-nav-label.mjs invocations above and check-undeclared-dep-imports.mjs).

All of the above were re-derived and re-run against the merged tree
(origin/main fetched and merged, no conflicts) at head 75749de5c.

Changeset

apps/docs is "private": true — this PR publishes nothing from any package. The
changeset gate's own counting logic (git diff --name-only --diff-filter=A $(git merge-base origin/main HEAD) HEAD -- '.changeset/*.md') was simulated locally
and returns 0 added changesets, so this PR needs the skip-changeset label — applied
via the additive labels endpoint right after opening, with a read-back to confirm.

---

Generated by Claude Code