Key changes:
- parallel (un)mount jobs via
concurrent.futures - more robust reachability checking:
- (nm): retry up to 5 times with 3s delay for mount
- (nm): unconditionally unmount (the interface is down so trying to reach a host causes more harm than good)
- (direct): run the checks once for mount and unmount
- command timeouts to avoid blocking
- unmount now tries to unmount gracefully first, after a timeout retries with
-fand after another timeout with-l -f - better function docs