Reset to use single OMP process in tests

[twsearle]

Description

OpenMP parallisation is leading to test result differences under the cray-with-intel-libraries build. It is unclear why this is happening, but seeing as we can't reproduce using the other builds and this build will not be available on the new HPC, I am removing the test for OMP for now.

Issue(s) addressed

Resolves #50

Impact

Fixes nightly test failures from orca-jedi.

Testing

• I have run mo-bundle to check integration with the rest of JEDI and run the unit tests under all environments

[matthewrmshin]

(Test was failing for spice_intel We don't do gnu_intel any more.)

[twsearle]

(Test was failing for spice_intel We don't do gnu_intel any more.)

Ahh OK sorry! I misread. In an earlier test this did pass so I think the PR will still be good (it was a temporary build that has been overwritten, hence the named build linked in the PR above) My understanding is that even the spice_intel build is not long term? Is that correct?

Regardless, I will make an issue to turn this testing back on in the future. Clearly something strange is happening. The openMP is rather simple, so it is puzzling it works under all other builds. It only gives us a few percent speedup, so it is not a concern to switch it off for now.

[matthewrmshin]

Looks good. Testing in my environment. Will approve when done.

[matthewrmshin]

Tested in my environment. Works as expected.