MYNN mods for ocean coupling and fractional mask

[benwgreen]

MYNN surface layer needs very slight changes if it will ever be used to compute surface heat/moisture fluxes (currently done by sfc_nst, sfc_ocean, sfc_sice, sfc_drv_ruc) in the fractional framework.
MYNN boundary layer needs an updated way to compute fluxes for the ocean in the case of fractional mask (this change follows the procedure done in GFS_PBL_generic.F90).
Testing finds these changes make no difference in the case of non-fractional mask and atmosphere-only, so regression test GSD_v0 should pass.
@joeolson42 is in favor of these changes.