fix(pm): the near-miss ownership census excludes seat posts and release announcements

The census (`ownershipMarkerNearMisses` over `OWNERSHIP_MARKER_NEAR_MISS_FORMS`)
listed five of #18914's sixteen rows as "no readable ownership record" where no
record is owed: four `pm:seat` posts, whose ownership is their BODY plus their
audit comments and whose comments carry shift narration, and one blocker-release
heading on a card nobody ever claimed.

Two exclusions, both COUNTED rather than dropped:

- `ownershipCensusSpeaksAbout` takes a `pm:seat` thread out of the POPULATION,
  reading the label off the page the sweep already holds (no extra request). The
  summary states the thread count and the line count, so the pre-filter census
  stays reconstructible.
- `releaseAnnouncementHeading` takes a release ANNOUNCEMENT heading out of the
  near-miss CHANNEL: a heading whose word is the participle `Released`, whose
  remainder names the `PR #n` that landed or the `pm:*` -> `pm:*` transition it
  unblocked, and which names no session. Three narrowings, each measured against
  the corpus line that forced it: heading forms only, participle only, no session
  token in either spelling.

Measured over objectstack's open board (539 open cards, 484 threads, tree
7843663): 36 near-miss lines on 30 cards before, 26 on 24 after; 9 lines on 5
`pm:seat` threads leave the population and 1 release announcement leaves the
channel. All ten of the other eleven rows that are still open survive
byte-identically.

The vocabulary is untouched and so is the reader: no form added, no change to
what a readable `Claim:` / `Release:` is.

Claude-Session: https://claude.ai/code/session_01BTeBejoPUvRHN8WdAJC6oF
Co-authored-by: Claude <noreply@anthropic.com>

Author: claude