Better handling of host key changes in remote filesystem sync

[ml-evs]

Sometimes the backup servers change host key, for whatever reason, and this makes them unable to sync. We should probably keep host key checking enabled, to protect ourselves from any man-in-the-middle attacks, but somehow it should be more automatic to rectify this issue when it occurs.

I would suggest a scheduled ping to the remote filesystems that checks their status and reports it at a healthcheck endpoint in the API. That way, diligent admins can decide to monitor the status and manually rectify where necessary. As it stands, this would require the integration of a scheduler such as APScheduler, as is also required for the scheduled backups (see discussion at #467).