Improve unreachable subsetting check

[simoncozens]

Description

This improves the com.google.fonts/check/metadata/unreachable_subsetting check in a number of ways:

• First, the condition font_codepoints was buggy! It searched through cmap tables to find those which are platform=Windows encoding=Unicode_BMP - which is fine if your font has codepoints exclusively within the BMP! I fixed it to use ttFont.getBestCmap(), which is simpler and also allows for cmap tables containing codepoints in supplementary planes.
• Second, we previously did not warn about codepoints which are subsetted out because they were not covered in any glyphset definition at all. This makes sense because it's not a font bug (it's a glyphsets one!) but it turns out that it's actually very useful to know how much of your font will just get dropped on the floor.
• Finally, this check only makes sense inside google/fonts, i.e. when you have a METADATA.pb file already. But it's also useful to be able to check during development (i.e. when you are upstream of google/fonts) if any codepoints will get left behind. We do this by copying the behaviour of the packager in how it detects subsets within a font, and using that information if a METADATA.pb file is not present.

Checklist

• update CHANGELOG.md
• wait for the tests to pass
• request a review

[simoncozens]

I'm not sure why you've assigned this back to me, and assigned it a milestone. It's ready to merge.

[felipesanches]

A bug in setuptools_scm caused this PR build to fail: pypa/setuptools-scm#905

Now that setuptools_scm version 8.0.0 was yanked, I'm re-triggering the build.

[felipesanches]

I'm not sure why you've assigned this back to me, and assigned it a milestone. It's ready to merge.

I always do it right before merging PRs and closing issues, with the purpose of tracking history of what went into each version of the program.