fix(subscriber): the disk fallback takes an audit status only when it is at least as new as the record

The audit mirror is written after the registry commits, so a redispatched
record (attempt 2, pending) can sit next to attempt 1's completed
status.json, and a failed mirror leaves it there. The fallback validated the
token and attempt against the record, which say nothing about the audit
status's provenance, and closed attempt 2 with attempt 1's completion
(review P1, Issue #57). A versioned record (transition_seq) now takes the
audit status only when its transition_seq is at least as new; an audit
status without a sequence cannot prove which attempt it describes and is
ignored by a versioned record. An unversioned record (never transitioned,
or MAM_ENGINE_CAS=0) and a missing record keep the old rule.

Regressions: the redispatch/flush interval, the failed-mirror leftover, the
unversioned audit status, a newer audit-only terminal, and conflict
refusals and legacy claims next to a moved registry.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

.agents/skills/lib_py/delegation/job_subscriber.py

@@ -84,11 +84,13 @@ def _check_disk_fallback(
         disk_status = None
         rec_token = None
         rec_attempt = None
+        rec_seq = None
         try:
             job_rec = load_job(jid, registry_dir)
             disk_status = job_rec.get("status")
             rec_token = job_rec.get("dispatch_token")
             rec_attempt = job_rec.get("attempt_id", job_rec.get("iteration"))
+            rec_seq = job_rec.get("transition_seq")
         except Exception:
             job_rec = {}
         if not disk_status or disk_status not in ("completed", "error", "cancelled"):
@@ -99,7 +101,20 @@ def _check_disk_fallback(
             except Exception:
                 logged = None
             if logged:
-                disk_status = logged.get("status")
+                # The audit mirror is written after the registry commits, so a
+                # record that is not terminal may sit next to the previous
+                # attempt's terminal status. A versioned record (the engine's
+                # transition_seq) takes the audit status only when the audit is
+                # at least as new; an audit status without a sequence cannot
+                # prove which attempt it describes, so a versioned record
+                # ignores it. An unversioned record (never transitioned, or
+                # MAM_ENGINE_CAS=0) and a missing record keep the old rule.
+                aud_seq = logged.get("transition_seq")
+                if rec_seq is not None and (aud_seq is None or int(aud_seq) < int(rec_seq)):
+                    logger.warning("disk fallback ignored audit status %r for %s: transition_seq %r is older "
+                                   "than the record's %r", logged.get("status"), jid, aud_seq, rec_seq)
+                else:
+                    disk_status = logged.get("status")
 
         if disk_status in ("completed", "error", "cancelled"):
             if expected_dispatch_tokens: