BUG: fix issues with wheel tags, and add more tests to cover tags

[FFY00]

Closes gh-141

[rgommers]

Thanks @FFY00. This looks like an improvement, it seems to fix things in the test case I just submitted in combination with Meson master.

One change is needed, to not error out on unknown platforms.

[rgommers]

Also, CI seems to be unhappy.

[rgommers]

More Mypy buggy-ness:

mesonpy/__init__.py:839: error: Returning Any from function declared to return "bool"  [no-any-return]

This works fine locally on both Linux and macOS, and the code is correct - just a little complicated with property decorators:(

[rgommers]

More Mypy buggy-ness:

It depends on the exact Mypy version and the minor version of Python ... sigh. On Python 3.9 this error disappears, and another one appears. Guess we have to live with it, Mypy isn't going to mature any time soon.

[rgommers]

This was all green modulo the mypy issue, the last commit should fix that.

[rgommers]

This is happy, in it goes. Thanks @FFY00!

[FFY00]

Thanks for all the work on getting this PR ready for merging. For next time, I'd prefer PRs that do multiple things to be rebase-merged, instead of squashed, the rule I try to abide by is one commit per logical change. In case you are not used to this kind of workflow, I recommend using fixup commits if you need to modify commits, as explained in https://openinput.readthedocs.io/en/latest/contributing/index.html#workflow. Thanks 😊

[rgommers]

Sure, that is a good idea in most cases. I am familiar with it, will aim to rebase-merge where possible.

In cases where it would break git bisect it may be a bit questionable though, then it's more of a judgement call and squash-merge or a merge commit may be more helpful.

[FFY00]

Thanks! All commits being self-contained is also something I strive to achieve, so in this workflow you would apply the fixes needed to the PR to each respective commit.