reconcile
Rescue SENDING rows stranded across a process restart. An event is marked SENDING just before transport; if the process dies mid-flight, the row stays SENDING forever and loadNextBatch never picks it up (it only selects PENDING/FAILED), so the event silently rots until retention deletes it.
staleBeforeEpochMs is an age gate so a concurrently-running process's genuinely in-flight rows aren't reset mid-send. Callers pass e.g. now - 2 x callTimeout.
Returns the number of rows reset.
Since
1.2.0