Deregister standbys after x mins of inactivity

[DAlperin]

Will fix to use replication slots later this is just a quick mvp

[DAlperin]

Tested by:

1. Setting up two node cluster: one primary, one standby
2. Set timeout to 1 minute instead of default 10
3. Kill standby
4. Verify standby no longer present in cluster after 1 minute
5. Restart standby and verify it cleanly rejoins cluster

I've decided to start writing down some test steps for myself to hopefully make writing integration tests easier eventually

[davissp14]

Just one note, but overall looks good!