dimension rescaling fixes and CFC cleanup

[klindsay28]

correct scale argument in get_param and chksum calls

add dimension scaling terms to some expressions

add some chksum calls

[gustavo-marques]

Thanks, @klindsay28. Just highlighting that, with the changes proposed in this PR, the workhorse CESM/MOM6 passes the unit scaling tests. We should discuss with @alperaltuntas how to incorporate these tests into one of the CESM Test suites.

[alperaltuntas]

...CESM/MOM6 passes the unit scaling tests

@gustavo-marques , when I set T_RESCALE_POWER = 11, I get answer changes in ocean.stats with GMOM_JRA.TL319_t061, even before merging this PR. Which configuration have you tested the unit scaling.

[klindsay28]

...CESM/MOM6 passes the unit scaling tests

@gustavo-marques , when I set T_RESCALE_POWER = 11, I get answer changes in ocean.stats with GMOM_JRA.TL319_t061, even before merging this PR. Which configuration have you tested the unit scaling.

@alperaltuntas , we indeed were not passing dimension rescaling tests prior to this PR. The first commit of the PR gets the tests passing for non-CFC runs. I verified this with each RESCALE_POWER parameter separately set to 20 and -20. I compared ocean.stats.nc, chksum output from DEBUG=True, and some history output that I had written out every timestep.

The other commits get the test to pass with the CFC tracers turned on, though you do need to set CFC11_IC_VAL and CFC12_IC_VAL to non-zero values. I checked this with each RESCALE_POWER parameter separately set to 20. I also checked the combination L_RESCALE_POWER=10,Z_RESCALE_POWER=11,T_RESCALE_POWER=13.

This testing was primarily with a GMOM experiment with the NUOPC driver. I also did non-CFC testing with MCT (CFC_cap is not implemented for MCT).

[alperaltuntas]

Thanks @klindsay28 for the explanation! I just confirmed that the PR fixes the scaling of my test case as well. I'll look into adding dimensional scaling tests to our aux_mom test suite. Not sure if there is a suitable test type tough. Seems like REP is the best candidate when combined with some hack I am thinking of. Will check with CSEG.

[klindsay28]

I'm definitely in favor of testing this in aux_mom.

A test that might serve as a useful example is CTSM's ciso test. It confirms that turning on carbon isotopes doesn't change answers.

I think it would be helpful if we could compare multiple runs to a base, preferably in a single batch submission. I wonder how hard it would be to generalize CIME's system_tests_compare_two to system_tests_compare_n.

[alperaltuntas]

Thanks @klindsay28 . Sounds like the CTSM's ciso example is a great place to start. I'll investigate.