Add tests for macOS key press simulation

[jugglinmike]

@ChrisC Here's a prototype of the system we discussed last week. I'm submitting it as a draft because there are three tests which don't work the way I expected them to work, and I'd like to try to understand why.

I documented them by including "(suspect behavior)" in their title and by skipping corresponding tests labeled "(intuitive behavior)". The "skipped test" approach surfaces my expectation in the tests' source code and in the output from executing them, which could be helpful if my hunch is correct and some of these behaviors are unavoidable.

We can't use this approach to observe press/release of the "Fn" key, but my initial testing suggested that we can't simulate pressing that key in the first place. This test suite will be especially reassuring if/when the code has to make the kinds of substitutions exemplified there (e.g. simulating "End" instead of "Fn + Right Arrow").

Resolves gh-12.

[jugglinmike]

I'm not too thrilled about this, but I'm somewhat less thrilled by the idea of defining driver globally.

[jugglinmike]

@ChrisC My research suggests that there aren't any practical workaround to the three oddities that these tests have uncovered. I've decided to simply document them; we still stand to gain a lot of insight even without addressing the underlying deficiencies.

(Also, it looks like I spoke too soon during our meeting today. The CI failures I observed were in fact transient and unrelated to the code itself.)

This is ready for review!