You might just kill the unhealthy instance. Cleanup its folders/paths (due to some contextual data that will remain there). And give a new instance with a new Raft ID to the cluster. With the flag --idx=4. It should work.
BTW, you have also to do /removeNode?id=3&group=3 if that doesn’t work. Try what I said.
As you have “renamed” the my flag. You might have to do this also the other instance.