Add a configurable timeout on the zerodeploy close() method

[balgillo]

We observed rpyc zerodeploy getting stuck on the close() method when running under Jenkins. Specifically, the following line from zerodeploy.py blocked indefinitely.

self.tun._session.proc.communicate()

This change adds a configurable timeout to the close() method, with default value 4 minutes, which is surely enough time to wait for anything to close. It makes it easier to debug issues when closing because it raises an exception with the stack trace, rather than blocking indefinitely with no logs printed.

[comrumino]

While your use case assumes 4 minutes is reasonable, the default behavior of the subprocess library favors a value of None. If you update the PR to specify None for the default timeout, then I will most likely merge your pull request. Of course, I would always welcome the addition of a unit test to validate the behavior but that is not a hard requirement :)

[comrumino]

While your use case assumes 4 minutes is reasonable, the default behavior of the subprocess library favors a value of None. If you update the PR to specify None for the default timeout, then I will most likely merge your pull request. Of course, I would always welcome the addition of a unit test to validate the behavior but that is not a hard requirement :)

[balgillo]

Thanks for taking a look at this @comrumino. I've changed the default back to no timeout, and I've added a unit test that checks that the timeout value is being passed through to the communicate() methods during close. Please merge this or let me know any other changes required.