storage: expose quota release queue and base index

This exposes the information on `_status/range/N`.

I've looked at a few ranges locally and I'm already noticing that
weirdly often the last index will be significantly ahead of the
quota release base index (though the release queue is always empty).

This might be something that only happens on single-node instances (the
release mechanism is based on Ready handling, which is different in
single instances because nothing ever needs to get acked externally),
but it's unexpected to me.

Motivated by cockroachdb#39022, where we see a proposal quota leak: quota is queued
up for release, but the base index is way ahead of the last index, so it
will never be released.

Release note: None

pkg/storage/replica.go

@@ -396,6 +396,9 @@ type Replica struct {
 		// replicas have persisted the corresponding entry into their logs.
 		proposalQuota *quotaPool
 
+		// The base index is the index up to (including) which quota was already
+		// released. That is, the first element in quotaReleaseQueue below is
+		// released as the base index moves up by one, etc.
 		proposalQuotaBaseIndex uint64
 
 		// Once the leader observes a proposal come 'out of Raft', we add the
@@ -908,6 +911,8 @@ func (r *Replica) State() storagepb.RangeInfo {
 	ri.NumDropped = uint64(r.mu.droppedMessages)
 	if r.mu.proposalQuota != nil {
 		ri.ApproximateProposalQuota = r.mu.proposalQuota.approximateQuota()
+		ri.ProposalQuotaBaseIndex = int64(r.mu.proposalQuotaBaseIndex)
+		ri.ProposalQuotaReleaseQueue = append([]int64(nil), r.mu.quotaReleaseQueue...)
 	}
 	ri.RangeMaxBytes = *r.mu.zone.RangeMaxBytes
 	if desc := ri.ReplicaState.Desc; desc != nil {