There used to be three version of average_forcing(), one for each
MOM_surface_forcing.F90. We moved two (solo/coupled) into
MOM_forcing_type.F90 and consolidated the diagnostics handles
into a type.

Note: have not applied finished process to the ice_solo_driver/
version of the model.

--Added diagnostic fields into fluxes type
--Added net surface heat flux diagnostic, including frazil and
  p-e+r.  This diagnostic compiles all terms that contribute
  to ocean surface heating.
--Changed name for MOM runoff_hflux to runoff_heat_content
--Changed name for MOM calving_hflux to calving_heat_content
--Left a hook for seaice_melt diagnostic, which requires changes
  to the coupler to access this term.
--Cleanup some indentation and some comments
--Diagnostics tested on global_ALE/z and GOLD_SIS
--Have yet to test on ice shelf driver, since there is no
  test case for that driver yet defined.
--We need to better merge the accumulation of surface
  boundary fluxes, including heat from runoff and calving.
  It is presently confusing, particularly the tempxPmE term.

- Allocation of fluxes%virt_precip was accidentally deleted
  for a specific configuration of ocean-only (benchmark)
- Problem manifest as segmentation fault.
- No answer changes.