Clear snapshots if process terminated in the middle of some procedure

[larisau]

It manager was stopped/killed during a backup, need to clear the snapshot from the disk as nodes may run out of disk space.
https://github.com/scylladb/siren/issues/6734#issuecomment-1201466898

[edit] TODO:

• on manager start, get the status of the cluster with Nodes UUID
• compare the retrieved Nodes UUIDs with UUIDs saved in backup_run (if there is any that can be resumed)
• if list of retrieved UUIDs is different than the ones stored in paused task then purge snapshot file and fail backup task

[karol-kokoszka]

After I read the thread from https://github.com/scylladb/siren/issues/6734 issue.
I understand this requirement as making SM able to recognize after startup if there are any backup tasks that are paused and were started on the different topology of the cluster. If there are any, then SM should:

• remove snapshot files from scylla data dir
• fail paused backup task with info that "topology changed"