fix(cli): retire pending rows a merge outdates; point identity rows at adjudicate

[jasonssdev]

Summary

Two commits.

#1266 — pending rows outlive merged concepts; purge drops the queue silently

• merge now retires as stale every open pending row that names the absorbed concept, including an identity row over a larger group and relation_type rows (new pq.retire_open_naming, called from queue_resolution.resolve_merged). Open rows that do not name the absorbed id are untouched; rows already applied, declined or stale keep their state. test_merge_leaves_an_identity_row_over_a_different_member_set_open asserted the old behavior and now asserts the row goes stale.
• purge's notice now says the findings.db drop also loses the whole pending-work queue and that openkos daemon --once rebuilds it; the jobs.db line now mentions the inbox watch history and that the next pass re-checks every inbox file. Purge stays warn-only.
• forget already swept the queue; unmerge is unchanged (the concept returns and the next pass recomputes its rows).

#1265 (pending side)

• Identity rows in pending now hint resolve: openkos adjudicate --apply (y merges, s skips, d records keep-distinct), with the keep-distinct command on an or: line; a group of more than two members is told that adjudicate --apply prints the pairwise merge commands. Rows already judged "same" keep their hint.
• A kind with 50 or more open rows gets note: 50 is the per-run candidate cap, so more may exist; ..., naming duplicates for identity and suggest-relations for relation_type. The queue records no truncation, so this is inferred from the row count: exactly 50 rows shows the note even when nothing was cut.

Specs: pending-work (merge-retire scenario, hint wording, cap-note requirement) and privacy-purge (the two notice lines); docs/cli.md pending and purge sections.

Related issue

Closes #1266
Refs #1265

Type of change

• feat — new feature
• fix — bug fix
• docs — documentation only
• refactor — no behavior change
• test — tests only
• chore / ci — tooling, build, or CI
• Breaking change

How was this tested?

• Observed RED before each fix: merge retire (('pending', None) == ('stale', 'stale')), the purge notice, 4 hint failures in test_pending.py, and the cap note.
• Mutation: dropping the open-status filter in the retire fails 3 tests (including one that keeps a declined row); >= → > in the cap check fails both cap tests.
• ruff check, ruff format --check, mypy ., pytest --cov (96.45%), evals/run_self_tests.py (46/46) pass locally.

Checklist

• My commits follow Conventional Commits.
• I added or updated tests for the change.
• I updated docs where behavior, interfaces, or the knowledge model changed.
• Lint, format, type check, and tests pass locally (ruff, mypy, pytest).
• Output remains OKF-conformant and derived stores stay reconstructible from the bundle + sources.
• The change is consistent with the project's guiding principles (local-first, provenance, freshness, human-in-the-loop).