Check if the compute capability of the host GPU matches the requested one

[bedroge]

Solves https://gitlab.com/eessi/support/-/work_items/257.

While I was working on this, I also found that there was a small issue with the way how multiple accelerator builds are done: each next one appends a --resume flag, as they keep adding that to BUILD_STEP_ARGS. That's solved now by resetting that variable for every iteration, and inside the loop the accelerator-specif flags are then added.

[casparvl]

Can you provide a way to test this PR? E.g. add an easystack for which the build was failing without this feature, and then prove that it works with this feature?

[casparvl]

Support meeting:
@bedroge initially had the issue on an older version of GROMACS, but that PR is already merged - so we can't use it for testing. We can try to trigger EESSI/software-layer#1524 on grace + CC70, that should then hopefully fail. If that's true, we can use the same easyconfig to test the fix here. Also, in the logs, it should be clear that the -nv flag is then only passed for native builds, and not for cross-compiled builds.

@bedroge : what if you're running a slurm job in which you request 1 GPU, but then in your job you see all? For now, it just checks the first one - that's probably the one you'd run on anyway.