RequestBatcher, which is used to batch and paginate intent resolution requests, has a number of parameters that affect performance (e.g. MaxMsgsPerBatch, MaxIdle, etc). However, these do not appear to have been fine-tuned for performance. We should set up a multi-node benchmark with realistic network latency for various scenarios (e.g. point intents vs ranged intents, differing intent density across the key space, etc) and see if we can improve intent resolution performance by tweaking the parameters.

Epic: CRDB-2554

Jira issue: CRDB-7131