Right now we do one index walk for batching + reverse lookup, and one for hashing. If we skipLookupForExtraKeys, then we still need to do the first index walk for batching and then another one for hashing. We should investigate refactoring/optimizing this and see if we can just do one. This would mean we need to move/add logic for checking batch limits into the hasher. We should wait until after SERVER-80682 which changes the way we do batch limits for extra keys check