Fail "rough pump stop" command on Edwards pump command failure.

The roughing pump at JHU somehow got confused, and is rejecting the stop command. Parse the return value and fail the MHS command.

cloomis@tron:rough1$ oneCmd.py --level=d rough1 pump stop
2020-11-11T16:57:45.887 sent rough1 pump stop (from cloomis_tron)
2020-11-11T16:57:45.892 rough1 d text="sending b'!C802 0"
2020-11-11T16:57:45.919 rough1 d text="received b'*C802 5"
2020-11-11T16:57:45.920 rough1 : text="5"

FWIW, in this case that 5 is for Invalid command in current state - e.g. serial command to start/stop when in parallel control mode, which is not literally true. But many strange and unusual things have been done to that pump and the associated MOXA in the last few days.