Gracefully shut down resharding state machines on stepdown/shutdown

The distributed algorithms underpinning resharding should be fine to pick up on another replica set node after a stepdown/shutdown. Accordingly, the only thing that needs to be done when one of these error types is thrown is to catch the error in an onError handler in the run function, log, and silently return.

It's possible that tenant migration has already solved this. Check in with their team/code.