These will be used to help extract metadata from wheels produced by CI
builds in order to publish them in PPAs linked to Artifactory.
I considered several options. `wheel-inspect` was a possibility, but it
has a number of additional dependencies, and it's not very clear which
versions of the metadata standards it supports. `distlib` looks
promising, but it seems to be a separate implementation of the packaging
standards that hasn't seen much adoption by other tools. I also
considered using the standard library's `importlib.metadata` or the
equivalent backport, but that only implements the actual `METADATA` file
format and leaves the problem of extracting it, so I was concerned that
there might be edge cases in how to do that which I was unaware of, or
that more might come up in the future.
`pkginfo` is used by the `twine` client to extract metadata before
uploading to PyPI, which seems like a strong recommendation in and of
itself, and it seems simple and well-maintained. It doesn't quite solve
the whole problem on its own: we also need to extract platform tags from
the wheel filename in order to work out whether a package is
architecture-independent. I could have implemented that with a regex,
but `wheel-filename` seems like a simple library encapsulating that sort
of thing so that we don't have to maintain it.