Better error handling for unexpected failures

[nolessafool]

Which problem is this feature request solving?

In situations such as the one handled in this slack thread netlify-only internal link, we had a lack of good visibility into failure causes in the situation where:

• the CLI exits before it really gets started (so DEBUG=* doesn't really say much that is useful)
• which was in this case caused by an unexpected signal 2/130 being received.

Describe the solution you'd like

It would be great if there were some better error handling during startup - in this case maybe even argument-parsing before it did any real work? Perhaps an explicit error handler for unexpected signals that aren't completely immediate (not sure what we can do about SIGSEGV for instance, but SIGINT is something that can be "caught" as far as I know) to show a stack trace or some other helpful debugging details? I'd guess we want it in a special flag but I'll leave those details to the experts who might build it.

Pull request (optional)

• I can submit a pull request.

[nolessafool]

cc @estephinson who worked on this situation