This commits adds a new field to MPAS-Ocean called
avgThermalForcing300m.  It is calculated as the thermal forcing
(difference between ocean temperature local freezing temperature) at 300
m depth.  It uses the temperature of the layer shallower than 300 m.
This could be replaced with vertical interpolation to get the value
exactly at 300 m.  The TF at 300 m is time averaged over the ocean coupling
interval.

This commit adds the So_tf300 coupler field for passing thermal forcing
at 300m through the coupler.  It also passes the avgThermalForcing300m
added to MPAS-Ocean in the previous commit to this coupler field, and
send it to the ismip6_2dThermalForcing field in MALI as the destination.

Note that a new list of coupler fields called x2g_tf_states_from_ocn
has been added to seq_flds_mod to differentiate this coupling field from
the iceshelf OCN/GLC coupling, which is handled differently.

This commit enables facemelting in MALI so that the new TF field will be
used to calculate a melt rate.  This is safe to enable in general,
because for situations where thermal forcing is not calculated  or not
applicable, it will be zero and facemelting will in turn be zero.  This
commit also updates the MALI output stream to write out the TF and
facemelting fields.

Note that this commit also changes the MALI output stream interval to
daily.  While that might not be the ideal long-term production solution,
it is likely the desired frequency for model development for the
forseeable future.

This commits defines the mapping for thermal forcing from ocean to glc.
It defines a special mapper for thermal forcing state (ocn2glc_tf_smap)
in CIME and MCT xml files.  It also declares and allocates mapper_So2g_tf
in prep_glc_mod but does not initialize or use it yet.

The mapping file requires some special treatment:
* it needs to use nearest neighbor mapping (which differs from most
  state remapping)
* it needs to include grid_imask in the ocean scrip file to only consider
  ocean cells valid if they are deeper than 300m

1. Differnetiate new ocn2glc TF coupling from existing ocn2glcshelf coupling
Create ocn_c2_glc logical in cime_comp_mod.  This controls the new
ocn2glc TF (thermal forcing) coupling separately from the existing
ocn2glcshelf coupling, which has its own ocn_c2_glcshelf flag already.
The new TF coupling will be active whenever ocn is present and glc is
prognostic.  These flags are used to initialize different mappers
independent of each other in prep_glc_init.  Those flags are also now
passed to prep_glc_calc_o2x_gx to control which mapping actually occurs.

2. Implement prep_glc_mrg_ocn
The existing ocn2glcshelf coupling handled a number of operations in
unusual ways because the fluxes themselves are calculated in the
coupler, so it was missing some of the standard operations for simply
passing fields between components.  As such, adding the new TF coupling
requires creating a prep_glc_mrg_ocn routine, which is responsible for
transferring fields from o2x_g to x2g_g arrays.  This routine was copied
closely from the existing prep_glc_mrg_lnd routine.

With this commit, the ocean TF field is successfully passed to MALI.

The entirety of existing ocn->glc coupling was for the ice-shelf
coupling.  To better differentiate the coupling in this branch based on
thermal forcing, this commit ensures there is either a 'shelf' or 'tf'
suffix on all ocn-glc coupling variables.

This allows testing with a MALI mesh where fjords are resolved.

The facemelting parameterization requires a subglacial runoff field as
an input.  Eventually, we may have this calculated prognostically in
E3SM from MALI and/or ELM, but that is not planned for the near-term.
Until then, a reasonable approximation is to use a constant historical
climatological field.

This commit updates the mpas.gis1to10kmR2 input file to include a
ismip6Runoff field provided by ISMIP6.  The field used is a 1995-2014
mean from the MIROC5 model, which was bias-corrected to match MAR over
that period.  See www.the-cryosphere.net/14/985/2020/ Fig. 2 and related
text for details.

This uses the layer that the desired depth is in, rather than the layer
above it.

Co-authored-by: Xylar Asay-Davis <[email protected]>

This commit adds an MPAS-Ocean namelist option for activating the
OCN-GLC TF coupling.  This option controls if E3SM should enable the
OCN-GLC TF coupling and also makes the associated TF calculations
conditional on the option being active.  The option defaults to false
and there currently are not any compsets that activate it, so it can
only be enabled with a namelist usermod at present.

Eventually, when we have compsets that require this mode, it should be
controlled analogously to MPASO_ISMF.

Four new gridspecs were introduced recently that were brought in to this
branch after a rebase and were missed in the original renaming of
OCN2GLC_*MAPNAME to OCN2GLC_SHELF_*FMAPNAME.  This commit makes the
name change for those grids as well.