fix(pm): convert the patrol's remaining count windows to announced time horizons (#13628)

H22's closed-card window and H23's commit window become TIME horizons with
announcing quota ceilings, following H8's conversion and H35's shape. H22's
divisor is named honestly: its rows are consumed by closed-issue UPDATE
activity (~188/day measured) rather than by closures (~11/day), which is why
its four-page cap covered a ragged 2.1 days of update-recency while reading
as a closure window — it missed 19 residue carriers closed within 3 days,
40 within 4, 265 within 7.

The three bare 10-page caps convert to the OTHER half of the shape. Measured:
none of them binds, and none is a recency window — each pages to exhaustion,
newest-created first. A time horizon there would discard the OLDEST open rows,
which is the population every row in this file exists for (an 8-day created
horizon drops 104 of 354 open cards; the oldest open card is 76.9 days old,
the oldest open P0 52.9). So they keep exhaustive paging and gain the
announcement they were actually missing.

What fell out as shared is the DISCLOSURE, not the paging: the divisors and
termination arguments genuinely differ, but "which of the two bounds ended
this pass" is one question in every case.


Claude-Session: https://claude.ai/code/session_01EXxTW8mvPBhoHxmyPZ63de

Co-authored-by: Claude <noreply@anthropic.com>

Author: claude[bot]