Because of SERVER-22131, the count command does not retry on replication errors. The code in shard_with_special_db_names.js should be changed to use itcount instead so it does not cause intermittent failures in the sharding continuous stepdown suite.