adds frazil (both the teos and basic options)

[alicebarthel]

This PR adds the basic and teos frazil option (latter documented in #462).

General implementation (documented in the docs):
The Frazil class calculates the frazil in a vertical column. We calculate the (cell, layer) tendencies and accumulate frazil fields that are necessary for frazil calculation in the layers above. The tendency class then simply calls frazil (if tendencies are turned on) and uses the tendencies. The column-integrated frazil fields (mass, energy, salt) will be passed to the coupler explicitly.

Teos frazil:
The frazil is thermodynamically consistent with the teos-10 ocean in omega. For now it relies on gsw submodules (cpu only) but will be updated once Eos features are added.
For conservation, we keep track of pure (fresh) ice mass, brine water mass, and brine salt mass separately within the ocean to allow melt to be energetically-consistent. It is fully conservative by construction.

Basic frazil :

• basicFrazil mirrors the mpas-o implementation logic into omega (non BFB due to eos and fundamental equation differences, it is a mass-based cousin of the mpas-o implementation);
• it runs on pm-cpu and pm-gpu so could be an option while we wait for the eos port that will enable the teos frazil option to run on gpus.

Caveats about the physics of basic frazil:

• it is non-conservative in energy by construction so would not recommend this for production runs. I manually changed the conservationCheck to be logging not failing to understand the behavior, but we probably want to switch it back for production runs with teos frazil that we know should be conservative.
• by default it uses a constant salinity of IceRefSal and is conservative in salt (and mass). There is a manual option to use porosity rather than constant IceRefSal. The simple ctest shows that this is very non-conservative in salt (order of magnitude difference!). The redistribution of the excess salt in the top layer reconciles this but is likely a strong surface forcing. I don't recommend this as a good approach either but we could later quantify this in single-column tests with more realistic conditions.

Scope and future work:

• we could easily make an intermediate version of frazil that uses constant latent heat yet tracks the thermodynamics more closely (e.g. tier 1: remove the enthalpy of melted equivalent in formation --> conservative. Tier 2: account for brine liquid mass and energy based on porosity). If the full teos frazil version is too expensive, we could consider this.
• @njeffery is developing a better mpaso frazil so we could update the omega basicFrazil to mirror it later.
• note: the "low salinity" check that was in mpaso has not been implemented here. A low salinity check will be implemented in the sea ice.

To do:

• turn off frazil tendencies by default to allow a comparison with baseline. Polaris tests with constant or linear eos need frazil turned off since frazil will fail without the teos eos.
• the conservationCheck currently uses LOG_INFO to allow testing on gpus (uses basic, thus non conservative). For production runs with teos frazil (when ready), conservationCheck should be changed back to LOG_ERROR to fail if non-conservative.

Checklist

• Documentation:
  • Design document is on overleaf (request access if you need to).
  • User's Guide has been updated
  • Developer's Guide has been updated
  • Documentation has been built locally and changes look as expected
• Linting
  • Pre-commit run locally
• Building
  • CMake build does not produce any new warnings from changes in this PR
• Testing
  • CTest unit tests (for basic option only) pass on pm-cpu and pm-gpu.
  • New tests:
    • CTest unit tests: added unit testing for basicFormation.
    • Polaris tests: TBD in single column.

[alicebarthel]

@andrewdnolan Let's chat about these terms so that we agree on the definitions and conversions for the SfcCoupling class.

[alicebarthel]

@andrewdnolan Let's chat about these terms so that we agree on the definitions and conversions for the SfcCoupling class.

[alicebarthel]

One key test is to run is the omega_pr with frazil tendencies turned off. This will give us confidence that we can merge this and leave frazil off if we have any concerns.
I did a rebase onto the current develop on a different branch (else omega_pr failed) and have a job in the queue to test that (55877874).

[cbegeman]

@alicebarthel It's my understanding that we could start testing this in single column configurations. Github is reporting some conflicts. Is it doable to rebase this onto develop?

[alicebarthel]

@cbegeman rebased on develop

[alicebarthel]

Suggested change

	FrazilTendencyEnable: true
	FrazilTendencyEnable: false

This needs to be off by default else the omega_pr tests will fail against baseline, and any test using constant/linear eos and frazil will fail. FYI @katsmith133 @cbegeman

[cbegeman]

Thanks, @alicebarthel !