SetupFlow should check if target peer's mirror_jobs table already has a mirror with the same name, even if catalog doesn't have metadata

[heavycrystal]

This can happen if the catalog database is truncated or otherwise modified, which could happen across installs. Also if multiple instances of PeerDB are running and they are connected to different catalog databases.

We could either error out, since we cannot be sure if another flow worker is running, or set sync_batch_id and normalize_batch_id to 0 and proceed with flow creation.