Use arm64 Linux runners for wheels

[radarhere]

GitHub has announced that arm64 Linux runners are now free for public repositories - https://github.blog/changelog/2025-01-16-linux-arm64-hosted-runners-now-available-for-free-in-public-repositories-public-preview/

This means that we can use them, rather than QEMU, to build aarch64 wheels, greatly reducing the time that it takes to build all of our wheels.

While I'm here, I've also added 'pypy' to CIBW_ENABLE, since it will be turned off by default in cibuildwheel 3. You can see a warning about this at https://github.com/python-pillow/Pillow/actions/runs/12755221825 and read more at https://cibuildwheel.pypa.io/en/stable/options/#enable

[hugovk]

Like python-pillow/docker-images#229, also very nice!

This PR replaces these 8 jobs, taking ~2+ hours each:

With these two, taking ~25 mins each:

It reduces total duration (wall clock) from ~2h 34m to ~45m, and total run time (total CPU time) from ~21h 24m to ~4h 30m! 🚀