Update continuous integration

[amontoison]

close #182.

[nimgould]

Just to mention that I tried ssids (galahad version) with the nvidia suite, and I can't compile one of the gpu codes (internal compiler error). However, compiling with cuda_nocuda.F90 instead of cuda.F90 (i.e, turn off GPU support) compiles and runs fine. Thus, for the time being, we should do thsi (and the archdefs system has been modified accordingly)

[amontoison]

Just to mention that I tried ssids (galahad version) with the nvidia suite, and I can't compile one of the gpu codes (internal compiler error). However, compiling with cuda_nocuda.F90 instead of cuda.F90 (i.e, turn off GPU support) compiles and runs fine. Thus, for the time being, we should do thsi (and the archdefs system has been modified accordingly)

I opened a PR in the GitHub repository of SPRAL to test it with nvidia compilers.
I have a similar compiler error than icpc...
ralna/spral#180

[amontoison]

I was able to compile GALAHAD (without SSIDS) with the Intel compilers on Mac.
All tests passed 🎉
I also fixed a bunch of other details.
On Windows, I need to dig more to understand the issue with the Intel linker.

[nimgould]

Very good. Do you know what the ssids issue is - what did the compiler grumble about?

[amontoison]

Internal compiler error 20000:
https://github.com/ralna/spral/actions/runs/7505963818/job/20436349072#step:10:335
We have this error with all versions of icpc (Mac, Linux), and icl on Windows.
The new compiler Intel icx on Windows that compiles both C and C++ files also has this error.