storage: re-enqueue Raft groups on paginated application

Fixes cockroachdb#31330.

This change re-enqueues Raft groups for processing immediately if they
still have more to do after a Raft ready iteration. This comes up in
practice when a Range has sufficient load to force Raft application
pagination. See cockroachdb#31330 for a discussion on the symptoms this can
cause.

Release note (bug fix): Fix bug where Raft followers could fall behind
leaders will entry application, causing stalls during splits.

Author: nvanbenschoten

pkg/storage/replica.go

@@ -3837,6 +3837,14 @@ func (r *Replica) handleRaftReadyRaftMuLocked(
 	const expl = "during advance"
 	if err := r.withRaftGroup(func(raftGroup *raft.RawNode) (bool, error) {
 		raftGroup.Advance(rd)
+
+		// If the Raft group still has more to process then we immediately
+		// re-enqueue it for another round of processing. This is possible if
+		// the group's committed entries were paginated due to size limitations
+		// and we didn't apply all of them in this pass.
+		if raftGroup.HasReady() {
+			r.store.enqueueRaftUpdateCheck(r.RangeID)
+		}
 		return true, nil
 	}); err != nil {
 		return stats, expl, errors.Wrap(err, expl)