add feature to override harikiri shutdown on verdi job execution

[pymonger]

Currently:

1. harikiri.py will see that no jobs came in for the last 10 minutes and flags the instance to terminate
2. harikiri.py does a random sleep to smear out thundering herd api calls to AWS
3. during the sleep, a job is queued that the verdi worker is pulling from and start processing that job
4. after waking from the sleep, harikiri.py gracefully shuts down all docker images and supervisord
5. The shut down of docker images is caught by verdi and thus results in a job-failed with exit code 143

Task is to add feature at start of each job to check for and cancel harikiri if it's in the middle of termination.