Our current targeted replica set tests for magic restore restore nodes to a new port, and then compare the data between source and destination. To do this, we have to manually modify replica set configs to ensure the new set can initialize. To more closely align with how Cloud performs restores, we can do a magic restore and start up the restored node on the original port. To perform the consistency checks, we can store the dbhashes from the source collections prior to restore, and then compare them with the destination dbhashes. We already do this for the sharded tests.

We should determine if we want to go back and modify the existing tests to follow this behavior. If we do decide to do this, we should ensure that both source and destination data files are preserved by resmoke in the event of a failure, to make debugging easier.