Refactor experiment.set_counter() and some minor bug fixes

[jo-basevi]

• Fix to payu crashing with empty restart files/dirs
• Add default values of LD_LIBRARY_PATHS
• Raise exceptions for missing model executables

Closes #285, closes #307, closes #306

[pep8speaks]

Hello @jo-basevi! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2023-10-03 23:26:51 UTC

[coveralls]

coverage: 42.053% (+0.04%) from 42.016% when pulling b5c0c30 on jo-basevi:285-fix-crash-with-restart into edf8d8c on payu-org:master.

[aidanheerdegen]

Looks great, thanks for cleaning that up and closing a bunch of issues.

Just a couple of tweaks and it should be good to go.

[jo-basevi]

A previous CI action failed at the lint stage because of payu/experiment.py:564:20: E1111: Assigning result of a function call, where the function has no return (assignment-from-no-return) where the code was in the payu run command:

for prof in self.profilers:
      if prof.runscript:
          model_prog = model_prog.append(prof.runscript)

I think it found a legitimate error so I fixed it in the previous commit (though could be wrong). But I am bit confused as to how that error isn't picked up when running pylint --extension-pkg-whitelist=netCDF4 --ignored-modules=backports -E payu locally or in other CI builds in other PRs..

[aidanheerdegen]

I think it found a legitimate error so I fixed it in the previous commit (though could be wrong). But I am bit confused as to how that error isn't picked up when running pylint --extension-pkg-whitelist=netCDF4 --ignored-modules=backports -E payu locally or in other CI builds in other PRs..

Yeah that is weird. It is a not often used part of the code, but you're right that a limiter should have picked that up before. Is computing supposed to be deterministic?

[jo-basevi]

Yeah that is weird. It is a not often used part of the code, but you're right that a limiter should have picked that up before. Is computing supposed to be deterministic?

Well it is now failing in another PR, so it'll be good to fix here..

[aidanheerdegen]

Good to merge I think. Thanks.