Some fixes for the new numpy 2.0 version.

[brownbaerchen]

See #442 for details. Still need to fix the parallel SDC project.

[pancetta]

Weird.. it looks like the parallelSDC tests are stuck..

[brownbaerchen]

Was that why you limited the matplotlib version? Is it plotting something weird?

[pancetta]

I don't recall exactly what the problem was, but there was one. Maybe tests got stuck..? 🙄

[pancetta]

I have no clue what's going on with those three tests. Can someone reproduce this locally?

[brownbaerchen]

No works for me locally. Where are even the tests for python versions newer than 3.10? Did they pass? I didn't find them.

[pancetta]

Huh, good question! That's more than weird.

[tlunet]

I've added more adaptations to numpy2 in the qmat-switch, the only problem left are for parallelSDC (test not passing).

In practice, removing the plotting function from the test make the test passe for 3 MPI threads, but not for 5 :
https://github.com/tlunet/pySDC/actions/runs/9625706930/job/26550873170#step:5:69

I suspect some additional multi-threading with the new versions of matplotlib (or numpy ...?) that make the mpirun execution reach a ressource limit that provoque the unending wait for the test

[pancetta]

Fixed by #445