Active/no-signal cleanup still leaves large review-only backlog after drain completes

[chubes4]

Problem

active-no-signal-drain --apply now safely removes many stale rows, but after completing every continuation it can still leave a large active_no_signal backlog with no higher-level next step beyond manual reports.

Evidence

After the improved drain completed with no next_commands, final hygiene still reported:

• active_no_signal: 107
• needs_metadata_reconcile: 35
• needs_full_review: 107
• suggested command: active-no-signal-report --limit=25 --offset=0 --format=json

The drain did its job for equivalent-clean / merged / remote-clean rows, but the operator experience still ends with a large review-only backlog and no compact triage buckets for why each remaining active/no-signal row could not be promoted.

Expected

Provide a compact post-drain summary that groups remaining active/no-signal rows by actionable reason, for example:

• remote branch still exists / PR open
• GitHub signal unavailable
• local branch not equivalent to base
• branch not merged remotely
• stale liveness but no safe cleanup proof

Ideally expose a next command that can continue bounded review by bucket, not just page through the raw report.

Notes

This is not asking to force-remove anything. The current safety behavior is good; the papercut is that the remaining backlog is hard to understand after the autonomous drain has done all safe work.

AI assistance

• AI assistance: Yes
• Tool(s): openai/gpt-5.5 via OpenCode
• Used for: Summarizing the remaining cleanup blocker shape and drafting the issue.