kill: Look for, and kill, the supervisor process

Useful if one wants to kill a job that is still waiting for its nodes to be
provisioned.

Signed-off-by: Zack Cerza <zack@cerza.org>

Merge pull request #2110 from Matan-B/wip-matanb-pexec-logs

teuthology/task/pexec.py: add logs to command executed

Reviewed-by: Aishwarya Mathuria <amathuri@redhat.com>

teuthology/task/pexec.py: add logs to command executed

The current output by pexec is:
```
INFO:teuthology.run_tasks:Running task pexec...
INFO:teuthology.task.pexec:Executing custom commands...
INFO:teuthology.task.pexec:Running commands on host ubuntu@smithi012.front.sepia.ceph.com
DEBUG:teuthology.orchestra.run.smithi012:> TESTDIR=/home/ubuntu/cephtest bash -s
```

The output should include the acutal command executed, similar to
exec.py:

```
INFO:teuthology.run_tasks:Running task exec...
INFO:teuthology.task.exec:Executing custom commands...
INFO:teuthology.task.exec:Running commands on role client.0 host ubuntu@smithi168.front.sepia.ceph.com
DEBUG:teuthology.orchestra.run.smithi168:> sudo TESTDIR=/home/ubuntu/cephtest bash -c 'sudo ceph osd pool create low_tier 4'
```

Signed-off-by: Matan Breizman <mbreizma@redhat.com>

Merge pull request #2107 from deepssin/fix-gevent-loopexit-dispatcher-crash

Fix dispatcher crash on gevent LoopExit exceptions