Update build system to use Meson

[tobiasdiez]

Following the example of Scipy/Numpy and sagemath/sage#36524, the build system is migrated to Meson.

I tried to stay as close as possible to the original autoconf implementation. A few differences are:

• No longer need to auto-generate signals.pxd, but have it at as a normal source file.
• Changed the authors to Sage Developers to align with the main sage repo. I'm unsure about this one and happy to revert it.
• Drop Python < 3.9 in alignment with https://scientific-python.org/specs/spec-0000/
• Remove the special handling when MINSIGSTKSZ_IS_CONSTANT (I had trouble making it compile and it's not constant for recent versions of glibc anyway)
• Remove the "ci-with-sage" jobs in the github action, which are non-working on the main branch and I've also no idea how to make them compatible with meson; I also think the current ci runs should give a pretty good coverage.
• Include the code of implementation_cygwin.c directly in implementation.c since I was not able to make it compile (it's also a bit hard for me to work on the cygwin code since I only have the ci to test it.)

Closes #179 (obsolete now)

[tobiasdiez]

Uff, tests are finally green.

[fchapoton]

Would you agree to engage yourself to maintain this for the years to come ?

[dimpase]

meson is a widely used and documented, unlike our present system to build this. besides, currently used distutils is deprecated, and needs a replacement anyway.

[fchapoton]

So we both agree that it's a good idea to merge, right ?

[tobiasdiez]

Would you agree to engage yourself to maintain this for the years to come ?

That's hard to promise but yes I'm willing to (help) maintaining it. The meson build info is also pretty straightforward, so it shouldn't be too hard for someone else to modify it.

Please merge #209 before this one here, and squash this PR to have a cleaner git history. Thanks!

[dimpase]

lgtm

[tobiasdiez]

Thanks @dimpase and @fchapoton!