[PSyData] Add/fix compilation tests for extraction

[hiker]

We need simple, stand-alone dummy libraries for gocean API and for LFRic, so we can add compilation tests (which need a mod file):

• [ ] Add dummy PSyData library for gocean API.
• [ ] Add a compilation test that real(kind=kind(1.0d0)) works as double precision.
• [ ] Add compilations tests for gocean extract case.
• [ ] Add dummy PSyData library for LFRic API.
• [ ] Add compilation tests for LFRic extraxtion (re-enable compilation in test_extract_kernel_and_builtin_dynamo0p3 and test_extract_colouring_omp_dynamo0p3)
• [ ] Make sure that e.g. the program name (which atm is the region name) is a valid Fortran identifier. Automatically created region names might be invalid (contain :, and are too long).

[arporter]

Do we need a dummy lib for gocean? The existing code is straightforward to compile and I think we are already in a position where we could get Travis to do this. I was going to open a new issue but then found this one...

[hiker]

This ticket was for extraction. While we do have a working code for gocean, but this is based on netcdf, and I thought we would better have a simpler library (e.g. one that only prints out the calls done and variable names). It would also be useful as a kind of template to develop other libraries for gocean. But if you think that the netcdf version is fine, then by all means use it ;)

[arporter]

Simpler is always good but given that you have a working netcdf version I think we might as well use it :-)

[hiker]

#1813 will provide the missing libraries, and allow compilation tests. I am keeping this ticket open to remember all the tests that need to be added.