When a container that a volume backup stopped no longer existed (for
example after a redeploy), recovery kept it as remaining. The execution
needed attention and every later run of the schedule was skipped.
- Treat "no such container" or "no such object" from docker inspect as
recovered; other failures still need attention
