SERVER-81974 introduced the execution of a setClusterParameter command as part of both addShard and removeShard, causing a slightly increased execution time of a user request when the cluster transitions from/into a single shard topology when it gets committed.

This overhead is now causing sporadic failures of the clusterwide_ops_with_add_remove_shards.js test, which heavily stresses this execution path and may go beyond the Evergreen timeout when run in slower variants.