SERVER-138505Open
Replicated fast count overcounts documents after transition to dedicated CSRS and back to embedded CSRS + restart
*Issue* The following sequence causes replicated fast count to go out of sync with the actual count and fail collection validation: # Start a sharded DSC sharded cluster with a embedded CSRS ("config shard"). # Create a sharded collection "coll" with 1 document on the CSRS. # Force a checkpoint and ensure fast count is flushed. # Transition the CSRS from embedded to dedicated. ** This moves all data out of the CSRS, and drops "coll" from the CSRS. # Transition the CSRS back from d