fix(bearings): gate unreclaimed completed work

[coreldh]

Intent

Republish PR 1735 cleanly from current main without rebasing or resurrecting its stale head. Close only the Bearings decision-hold regression: a terminal-looking but unreclaimed in-flight non-held task whose endpoint and worker are both verified absent must surface as a measured-liveness gate; held, active-worker, and live-blocked lanes must not receive that gate. Prove RED before GREEN and prove the property goes RED for the intended reason when the measured-liveness selector is neutralized. Keep the replacement tight, preserve production semantics, and do not absorb the oversized-fleet snapshot/argv work from the separately published PRs 1968 or 1964. Publish only through this repository's no-mistakes pipeline; do not force-push, merge, self-approve, or close PR 1735.

What Changed

• Added measured-liveness data to fleet snapshots and gated terminal-looking, unreclaimed non-held tasks when both endpoint and worker are verified absent.
• Preserved held, active-worker, and live-blocked behavior while expanding regression coverage, including selector-neutralization and teardown cases.
• Documented the Bearings gate behavior across the skill, architecture, and decision-hold lifecycle guidance.

Risk Assessment

✅ Low: The change is tightly scoped to Bearings liveness measurement, requires verified absence of both endpoint and worker, preserves held and active lanes, and includes RED-before-GREEN falsification coverage without absorbing unrelated fleet work.

Testing

The targeted lifecycle test passed and produced reviewer-visible CLI evidence. No UI surface applies.

Evidence: Bearings regression transcript

Focused regression passed; the transcript includes the measured-liveness gate, negative lanes, and selector-neutralization RED check.

FM_TEST_BEGIN 2026-08-09T02:11:39Z tests/fm-decision-hold-lifecycle.test.sh family=pure-contract-unit expected_gate_skip=none
ok - verified-absent in-flight work gates while held and active workers avoid measured-liveness gates, selector neutralization goes RED, and completion refuses before report loss
ok - non-forced scout teardown always requires durable inventory verification
ok - captain holds are idempotent, distinct, teardown-safe, Bearings-visible, and durably routed before close
ok - completion and verification validate origins before constructing paths
ok - ended visual review follows the same decision-hold completion owner
ok - resolved findings and decision-like prose do not create false holds
ok - terminal single-owner stale status decisions do not block empty inventory
ok - main-home and secondmate-home captain holds remain correctly routed
ok - resolve matches first/middle/last in quoted blocked_by and rejects a genuinely absent id
FM_TEST_END 2026-08-09T02:12:37Z tests/fm-decision-hold-lifecycle.test.sh exit=0 duration_ms=57992 gate_skip=false
FM_TEST_SUMMARY total=1 failed=0 skipped_gate=0 duration_ms=58078
FM_TEST_SUMMARY_FAMILY family=pure-contract-unit count=1 duration_ms=57992 failed=0
FM_TEST_SLOWEST rank=1 script=tests/fm-decision-hold-lifecycle.test.sh duration_ms=57992

Pipeline

Updates from git push no-mistakes

✅ **intent** - passed

✅ No issues found.

✅ **Rebase** - passed

✅ No issues found.

✅ **Review** - passed

✅ No issues found.

✅ **Test** - passed

✅ No issues found.

• bin/fm-test-run.sh tests/fm-decision-hold-lifecycle.test.sh
• Verified phantom, held, active-worker, selector-neutralization, and teardown-preservation cases.

✅ **Document** - passed

✅ No issues found.

✅ **Lint** - passed

✅ No issues found.

✅ **Push** - passed

✅ No issues found.