Executing protocol loses work and ties up consumer when view changes

When a run is submitted to another cluster member and that member leaves the run is re-requested, but from the wrong owner. This results in the run never being completed and an awaiting consumer being removed from the coordinator list indefinitely.