sagemath: update to 10.3.

[tornaria]

PR for tracking sagemath beta/rc until 10.3 is released.

Testing the changes

• I tested the changes in this PR: briefly

[tornaria]

@ahesford testing here cython 3.0.7 to compile and check sagemath

[ahesford]

Thanks for checking Cython; updated in 0dea09d.

[tornaria]

@dkwo sagemath 10.3 is close to being released. This PR is up-to-date with rc1, using flintlib 3.1 (it's just simpler to update both at the same time -- singular needs patching as well).

Also I changed (slightly) the package, first to use pep517 build (which works fine) and also moved the internal scripts from /usr/lib/sagemath/10.3/bin directly into /usr/libexec/sagemath. Both changes should be transparent to the user.

If you want to test (esp. in aarch64) let me know of any trouble. We have a small window to contribute fixes upstream if something pops up.

[dkwo]

the changes look good. it's failing at building givaro though, both on aarch64 and x86_64: ./givaro/givdegree.h:42:33: error: 'uint64_t' does not name a type

[dkwo]

maybe we need this linbox-team/givaro@a6b3708 ?

[tornaria]

I added the patch to givaro; revbump to be safe since the missing include is in a header that we ship in givaro-devel.

[tornaria]

I'm getting weird "error" in CI (no log at all), I don't know what is going on...

[tornaria]

It turns out that the flintlib-3.1.0.tar.gz tarball had changed (to fix two issues I reported, see flintlib/flint#1809) hence the template was failing at fetching it. It's updated now.

[tornaria]

I'm getting weird "error" in CI (no log at all), I don't know what is going on...

It seems the logs are now XHRed from a *.windows.net address instead of *.github.com, which umatrix was blocking. This confused me as it seemed the CI itself was completely broken.

[dkwo]

It all builds and passes checks on aarch64 now, thanks!

[tornaria]

@dkwo final release of sagemath 10.3, if you want to review and give one more try before we merge.

Notes

• gap: I'm holding back 4.13.0 since there is no PR available on sagemath yet (Gap 4.13 released, some doctest failures sagemath/sage#37616)
• singular: updated to 4.3.2p16 for flint 3.1 support, needs a small make sagelib work with singular>=4.3.2.p15 (future 4.4) sagemath/sage#37492; I added myself as co-maintainer
• switched to pep517 build, but still from a github tarball; to build from pypi sdist we are missing sagemath-standard: include sage_setup in sdist sagemath/sage#37287

[dkwo]

This builds and checks fine on aarch64 :)
time to merge it? @leahneukirchen

[leahneukirchen]

Then mark as ready.

[tornaria]

Ready to merge! Thanks @dkwo @leahneukirchen for your support.