The current testing setup for transition_to_dedicated_config_server.js creates one sharded collection and one unsharded collection with the config server being the DB primary. However, during a config server stepdown, insertions on the unsharded collection could trigger a writeConcernError due to InterruptedDueToReplStateChange.

As part of this ticket, we should enforce retryWrite: true so that the operation is retried in the case of a stepdown.