In SERVER-79734 we added a check in ReplSetTest.startSet when restart=true that if the caller has not disabled auto-step up behavior but they are using the in-memory storage engine we throw an error. (Without this check, the user would see errors from the subsequent call to replSetStepUp noting the replica set is not yet initialized.)

BF-32585 has shown this serverStatus call can race with sharding initialization, however it's not trivial to make this call un-racy in a robust way.

For now I think we should remove the check but we can reevaluate if we see a lot of people being confused by the need to set triggerStepUp=false when restarting in a test with the in-memory storage engine in the future.