Update to CMEPS 1.0.25

[blimlim]

This PR forms part of the work for ACCESS-NRI/dev_coupling#51

It merges our version of CMEPS, based on version 0.14.35, with CMEPS1.0.25, which is the version currently used in OM3.

I'm using a duplicate access-cmeps-0.14.35-merge-copy branch because resolving the merge conflicts applies a merge commit to the feature branch, and I'm hoping to not break our original access-cmeps-0.14.35 branch. I'll add separate comments elaborating on the choices for the different merge conflicts.

[blimlim]

Notes on the merge conflicts:

1. med_history_mod.f90:

<<<<<<< access-cmeps-0.14.35
                  use_float=.false., rc=rc)
=======
                  use_float=.true., ntile=ntile, rc=rc)
>>>>>>> access-cmeps-1.0.25

Here I used the access-cmeps-1.0.25 change. Attempting to keep use_float=.false along with the ntile changes led to the issues discussed here

2. med_phases_prep_atm.f90:

<<<<<<< access-cmeps-0.14.35
    if (trim(coupling_mode) == 'cesm' .or. &
        trim(coupling_mode) == 'nems_frac_aoflux' .or. &
        trim(coupling_mode) == 'hafs') then
       call med_merge_auto(&
            is_local%wrap%med_coupling_active(:,compatm), &
            is_local%wrap%FBExp(compatm), &
            is_local%wrap%FBFrac(compatm), &
            is_local%wrap%FBImp(:,compatm), &
            fldList, &
            FBMed1=is_local%wrap%FBMed_ocnalb_a, &
            FBMed2=is_local%wrap%FBMed_aoflux_a, rc=rc)
       if (ChkErr(rc,__LINE__,u_FILE_u)) return
    else if (trim(coupling_mode) == 'nems_frac' .or. &
             trim(coupling_mode) == 'nems_orig' .or. &
             trim(coupling_mode) == 'nems_frac_aoflux_sbs' .or. &
             trim(coupling_mode) == 'access') then
       call med_merge_auto(&
            is_local%wrap%med_coupling_active(:,compatm), &
            is_local%wrap%FBExp(compatm), &
            is_local%wrap%FBFrac(compatm), &
            is_local%wrap%FBImp(:,compatm), &
            fldList, &
            rc=rc)
       if (ChkErr(rc,__LINE__,u_FILE_u)) return
    end if
=======
    call med_merge_auto(&
         is_local%wrap%med_coupling_active(:,compatm), &
         is_local%wrap%FBExp(compatm), &
         is_local%wrap%FBFrac(compatm), &
         is_local%wrap%FBImp(:,compatm), &
         fldList, &
         FBMed1=is_local%wrap%FBMed_ocnalb_a, &
         FBMed2=is_local%wrap%FBMed_aoflux_a, rc=rc)
    if (ChkErr(rc,__LINE__,u_FILE_u)) return
>>>>>>> access-cmeps-1.0.25

Again I used the change from access-cmeps-1.0.25. This PR dropped the separate treatment for different models in these merges, instead including the albedo and flux field bundles for every model. Discussing this with @kieranricardo, it sounds like using the access-cmeps-1.0.25 changes should be ok here as the extra unused field bundles shouldn't have any impact.

There's an analogous conflict in med_phases_prep_ocn_mod.f90 but the diff ended up a bit broken. I followed the access-cmeps-1.0.25 changes here too.

[kieranricardo]

Looks good - nice work getting this running!

Just some old comments and logging of mine that I think we can delete now

[blimlim]

Thanks @kieranricardo! Merging now