fix: if a watch closes and fails to reconnect, exit the operator

[secondsun]

This is one of my proposed fixes for #395

[secondsun]

@metacosm I do get some different intermittent failures, but the failing test here is working locally. Thoughts?

[wtrocki]

@secondsun we need this test to adjust to our solution:

DefaultEventHandlerTest.executesTheControllerInstantlyAfterErrorIfEventsBuffered:150 

After that build should pass and that would be easy merge

[metacosm]

LGTM

[adam-sandor]

I think this is a good solution for a rare event. I like that the framework gets tested on some buggy clusters so we can get these edge cases handled.

[metacosm]

Going ahead with the merge despite the tests failing. I've checked the PR locally on my system and everything works fine. Not sure why the tests are failing all the sudden.

[secondsun]

The tests also passed before the rebase so they work sometime!