Forward reads to primary if replica is behind a lot

[penberg]

If a replica suspends itself because of inactivity, there might have been lots of writes to primary until the replica resumes. Let's look into ways to initiate replication in the background, but forward reads to the primary until the replica has caught up.

[MarinPostma]

The primary could send its replication index as part of the initial handshake, based on that we can decide whether to proxy or not