Mount can remain stale after transient HEAD reconciliation failure

A transient failure while resolving HEAD or publishing its snapshot consumes the watcher event without retrying, leaving the mount on the old commit until HEAD changes again or the daemon restarts.

Start by tracing the watcher event handling for HEAD reconciliation and snapshot publication. Reproduce or test a transient failure in either step, then verify that the event is retried and the mount eventually reflects the new commit without requiring another HEAD change or daemon restart.

Written by the indexing model from the issue text.