improve signal handling and worker cleanup

[epwalsh]

Fixes #5369. @dhruvdcoder could you please take a look and verify that this fixes your issues?

Changes proposed in this pull request:

• allennlp will now catch SIGTERM signals and handle them the same way of SIGINT (keyboard interrupt).
• The MultiProcessDataLoader will properly shutdown its workers when a SIGTERM is received.
• This also adds an additional worker safety mechanism that allows MultiProcessDataLoader workers to constantly check if the main process is still alive and consuming. If they determine that the main process has exited, they can shut themselves down. This is useful if the main process exits abnormally. But unfortunately it only works when start_method is "spawn".

Before submitting

• I've read and followed all steps in the Making a pull request
  section of the CONTRIBUTING docs.
• I've updated or added any relevant docstrings following the syntax described in the
  Writing docstrings section of the CONTRIBUTING docs.
• If this PR fixes a bug, I've added a test that will fail without my fix.
• If this PR adds a new feature, I've added tests that sufficiently cover my new functionality.

After submitting

• All GitHub Actions jobs for my pull request have passed.
• codecov/patch reports high test coverage (at least 90%).
  You can find this under the "Actions" tab of the pull request once the other checks have finished.

[dhruvdcoder]

@epwalsh Thank you for the quick fix. I will incorporate this in my full sweep setup today and get back to you.

[dirkgr]

This seems OK, though I'm wondering if you need two connections (the queue and tx). Can't you send a terminal value through the queue to tell workers to stop?

[epwalsh]

The existing queue moves data from the workers to the main process, not the other way around, so there needs to be a separate channel of some kind to move data in the opposite direction.