*Implement sub-cycling for internal tides refraction/propagation

[raphaeldussin]

This PR allows to bring back the thermodynamics timestep higher while keeping the raytracing under CFL by doing multiple steps within the thermo step.

The only variable that needs to be accumulated in the subcycling is the slope loss (residual of transmission and reflection on slopes) Since it is already accumulated within the 2 directions of propagation, it only needs to be initialized outside of the subcycling.

[raphaeldussin]

my quick and dirty testing (2 days of OM4) shows we were able to come back to DT_THERM = 7200 from 3600 with INTERNAL_TIDE_SUBCYCLING = 2 and the timing for the "Ocean thermodynamics and tracers" is down from 57s to 32s (consistent with doing only half the work) compared to 28s for a run without the parameterization and DT_THERM = 7200

[raphaeldussin]

I was able to reproduce previous answers with DT_THERM = DT_ITIDES = 3600 and reproduces across processor counts.

[Hallberg-NOAA]

These changes now look good to me. Thank you for the addition of this valuable new capability that will make the internal tides code much more efficient and make it practical to
use them in GFDL's Earth system models.

[Hallberg-NOAA]

This PR has passed pipeline testing at https://gitlab.gfdl.noaa.gov/ogrp/mom6ci/MOM6/-/pipelines/27743. I would have expected there to be changes in runtime parameters, but there were not, which is telling us that we probably need to add the internal tides to one of the standard pipeline tests to give us coverage of this important new capability.