Add MOM_generic_tracer to ocean_BGC

[theresa-cordero]

@yichengt900

This PR is in response to issue #36.

Following PR #790 to MOM6 dev/gfdl, the MOM_generic_tracer.F90 file has been removed from MOM6/src/tracer. A copy of MOM_generic_tracer.F90 is now needed with the ocean_BGC code. This commit copies MOM_generic_tracer.F90 "as is" from MOM6.

Users of more recent versions of MOM6 (commit fcf5fff or later) with ocean_BGC need this file in ocean_BGC or to dramatically change their compile instructions. We recommend tagging the versions of ocean_BGC as:

• ocean_BGC.v2024: without MOM_generic_tracer.F90, compatible with MOM6 before commit fcf5fff
• ocean_BGC.v2025: with MOM_generic_tracer.F90, compatible with MOM6 commit fcf5fff or later.

[yichengt900]

@theresa-cordero, thanks! I’ve tested this draft PR with the latest gfdl/dev MOM6 branch. The coupled model built successfully, and I was able to run both a 2-day 1-D case and a 1-degree global model without any technical issues. (PR #34 will be required if someone forgets to turn on internal_heat in MOM6, though it is enabled by default in most OM4 configurations.)

[theresa-cordero]

Thanks for testing this Yi-Cheng. We discussed adding more information to the README but there doesn't appear to be one. I think a README would be helpful especially as we move to a version 2 and version 3.

[theresa-cordero]

Should we consider moving this file to a subdirectory? something like MOM6_interface or even gfdl_interface in order to make it easier for other groups to point to their own version during compile time.
@dougiesquire would that be better when using with WOMBAT?

[yichengt900]

@theresa-cordero, I support adding a subfolder for the interface files. I think gfdl_interface might be a better choice, but I’d like to hear others’ opinions.

For the README, we can include short instructions on how to compile ocean_BGC with MOM6, both before and after this commit. I recall @nikizadehgfdl offered to create a release tag—perhaps for the version before this PR. This would allow users to refer to that tag when compiling ocean_BGC with the older MOM6.

[dougiesquire]

I don't think it really matters too much to us, thanks. As we discussed, we (ACCESS) need to maintain the "genericness" of generic tracers so I think we will continue to use in our fork MOM_generic_tracer as it currently exists.

What would be really helpful is that there is clear documentation once you start to commit code that isn't generic. E.g. my understanding is that COBALT v3 will require a specific version of MOM_generic_tracer and won't work with MOM5.

[dougiesquire]

MOM_generic_tracer as it currently exists.

Sorry this is unclear. By "as it currently exists", I mean the unmodified version taken directly from MOM6 that this PR adds.

[nikizadehgfdl]

@theresa-cordero are you done with this PR? Can I merge it? Thanks.

[theresa-cordero]

@nikizadehgfdl I think this PR is good to be merged

[theresa-cordero]

@nikizadehgfdl A readme has been added to go with this PR in PR #38