Snapshots taken while backup agents are running cause bad behaviour when restored

Regular backups do not restore the /xff keyspace, while snapshots do. The consequence of this is that the cluster restored by a snapshot will have its backup (and DR) ranges filled in (located in the /xff keyspace), which are present when a backup agent is active. If a backup agent is then started in the new cluster, it will read from the backup ranges and spawn tasks according to the old data. This can cause corruption of data and/or crashes in the new cluster.