In generic_tracer_source call, perform unit conversion for internal heat only when it is applicable

[yichengt900]

This PR is related to another PR made in the NOAA-GFDL/ocean_BGC repo.

In the past, setting DO_GEOTHERMAL = True was necessary for COBALT to function correctly. Failure to enable geothermal heating will lead to segmentation faults if someone turned it off accidently.

We have now made internal_heat an optional argument in the generic_COBALT_update_from_source. The if(present()) check will determine if internal_heating is a null pointer. However, in the unit conversion case, the scaling operation G%US%RZ_to_kg_m2*US%C_to_degC*tv%internal_heat(:,:) occurs before passing the data to generic_tracer_source. This will cause a segmentation fault if internal_heat is null.

This PR (proposed by @andrew-c-ross originally) fixes this issue by omitting this argument if internal_heat is not associated.

[Hallberg-NOAA]

I agree that these changes correct the problem that is being described.

In the longer, term, it might be even better if we could arrange to (optionally) pass a MOM unit_scale_type argument into generic_tracer_sources(). This would have the benefits of limiting the code-duplication in MOM_generic_tracer.F90 while also allowing us to propagate the dimensional consistency testing for physical variables into the generic tracer code.

[adcroft]

@yichengt900 Can you check "Allow edits and access to secrets by maintainers" at the bottom of the right hand column (also when you submit a PR)? Our branch rules block merges for out of date code w.r.t. dev/gfdl. That option allows us to update the branch.

[yichengt900]

Thanks, @adcroft. I think Marshall pointed it out the same issue, and it appears that the option Allow edits and access to secrets by maintainers does not show up when merging an organization fork (weird). I have manually merged the recent updates into my branch.

[adcroft]

Testing at https://gitlab.gfdl.noaa.gov/ogrp/mom6ci/MOM6/-/pipelines/24226