Hard to collect stack for hanging tests when run in parallel

[vito]

We often run tests with -p or -nodes=X (both in CI and during dev). Unfortunately this makes it impossible to get a stack dump when the tests are hanging. If you send SIGQUIT to the toplevel ginkgo process you just get a stack dump of the coordinating node, not the nodes actually running the tests. If you send SIGQUIT to the nodes running the tests, the coordinating node just prints:

	 -------------------------------------------------------------------
	|                                                                   |
	|  Ginkgo timed out waiting for all parallel nodes to report back!  |
	|                                                                   |
	 -------------------------------------------------------------------

It'd be super handy if ginkgo trapped SIGQUIT, dispatched it to each of its nodes, collected their output, and then printed them non-interleaved. It should probably also do its own stack dump + exit, in case things are stuck there.

[onsi]

Yes. Trapping and retransmitting would be the way to go here. I believe it already does this for SIGINT.

.......PR?

[MissingRoberto]

Same here. It's really tricky to find out the hanging test.

[onsi]

Interrupts are handled much more cleanly now in v2 - sending SIGINT should get you a much clearer sense of where things are hanging.