After chunk migration, the donor shard deletes orphan documents (the documents that have been copied to the recipient). We should track these in query stats since it could be a big part of the customer workload and there are not a lot of Atlas metrics that make sense of it. 

Including chunk migration process here since I found it helpful from glean (trying to move chunk from donor shard to recipient shard)

1. recipient shard clones base docs from donor shard (these are inserts), donor tracks concurrent deletes/inserts/updates that occur on the migrating range as transfer mods. (This part should be tracked in query stats already)
2. donor shard waits for recipient to catch up on transfer mods
3. commits the docs and orphan cleanup happens - donor's local copies of documents in the migrated chunk range are deleted (not tracked in query stats)