Investigate performance of SCP message capturing in database as validator set grows

[bboston7]

The current tier 1 validator set is ~23 nodes, but is expected to grow 2-3x. As of right now, PR #4121 updates the set of SCP messages for a given ledger by deleting the database entries corresponding to that ledger and then re-adding all messages to the database. We do this because the alternative of tracking deltas and only updating those rows in the database is slower. As the validator set grows this performance characteristic may change. Once the network has more validators, we should re-evaluate this technique and consider performance gains from tracking deltas and/or moving this work to a separate thread.

[bboston7]

I took a look at Grafana around when this change landed and didn't see any noticeable performance changes. Most crucially, the average apply time is unaffected, which is what we were primarily concerned about. Given all that, I'm closing this ticket.