Pass seaice_melt field through ice-ocean coupler

[hdrake]

Presently, the MOM6 diagnostic for liquid precipitation into the ocean (prlq) includes both the expected contribution (IOB%lprec_top_ocn in SIS2) and, for legacy convenience, also the net sea ice melt/formation (local net_melt field in SIS2). By bundling these two together, it is impossible to currently separate the surface mass fluxes from ice-ocean interactions from the hydrological cycle (without additional processing using some ice diagnostics).

This PR would instead separate the two fluxes, renaming net_melt to IOB%seaice_melt in SIS2 and passing this on to fluxes%seaice_melt in MOM6 via the FMScoupler. The fluxes%seaice_melt field in MOM6 corresponds to CMOR diagnostic fsitherm, which Griffies et al. (2016) identify as a level 2 priority diagnostic in the OMIP protocol for contributing to CMIP6. To my knowledge, this is the only diagnostic that is currently preventing MOM6 from being consistent with those protocols.

It should be a priority to merge this PR as soon as possible so these fields are available for OM5 and ESM4.5 development (and production runs).

See companion PRs to NOAA-GFDL/SIS2#214 and NOAA-GFDL/MOM6#710

[rem1776]

@hdrake Would you be able to merge in the main branch to this PR? Looks like there's some conflicts that came up since this was put in.