When verifying that test harness's SSH and SOCKS servers have been

started check also that the process is actually alive, since they
could have died once the pidfile was written out