Description of problem:
Upstream has observed a timing condition where the shadow can hang and the job stays idle during a transition at the end of the job. 

Version-Release number of selected component (if applicable):
condor-7.6.5-0.6

How reproducible:
100%

Steps to Reproduce:
In the configuration of the execute machine:

   MachineMaxVacateTime = 0
   KILLING_TIMEOUT = 0

In the submit file:

   periodic_remove = JobStatus == 2
  
Actual results:
When the shadow deactivates the claim, the startd hard-kills the starter, so there is no final update. The shadow then waits around indefinitely for a final update that will never arrive.

Expected results:
Clean Exit of job and correct update of JobStatus

Additional info:
https://condor-wiki.cs.wisc.edu/index.cgi/tktview?tn=2591