changelogtemptochangelog make configurable number of entries to process at once

As discussed in tier-grouper slack channel:
Since the changelogtemptochangelog processor is singular and exclusive in what it does… why not grab 500 or 1000 entries at a time, map and then insert in one statement or some other more efficient manner and then delete in bulk? It seems all the popular DBs would support this rather well and should dramatically increase the throughput of this process. Of course, make transactional where possible to limit replay. Default to 1 (like now) or make it 10, preferable. Should have significant impact on throughput of grouper changes.