Geo: Actively try to correct stale checksum on the primary

Summary

With !5196 (merged) the secondary will actively try to correct verification failures marking the repositories to be resynced, but for repositories that have stale checksums in the database no matter how many times the secondary resyncing them because resyncing does not change the checksum in the database.

Right now if we don't detect stale checksums in the primary database, we should figure out how to discover them and trigger a repository updated event when it is fixed.

Assignee Loading
Time tracking Loading