Sync with cesm2 1 x rel

[mnlevy1981]

Description of changes:

Brings the cesm2_1_x_rel branch onto master -- done to get OMIP compsets and updated JRA forcing support. Note that this branch requires modifications in CIME as well (will link to that PR when it is ready)

Testing:

Test case/suite: aux_pop on cheyenne (intel compiler only), aux_pop_MARBL on cheyenne (intel and gnu) and hobart (nag and pgi) Test status: climate changing in C & G compsets (except OCN_GRID=gx3v7, CORE normal year forcing, and no ecosys)

1. sfwf_filename changed for CORE forced gx1v7 runs, which changes answers for all runs of this resolution
2. gx3v7 BGC tracer initial condition files changed, which changes answers for all runs with ecosys_on = .true.

I compared the aux_pop results to a baseline generated with the cesm_pop_2_1_20200318 tag (and also cime5.8.17 CIME tag). Those results are on glade:

/glade/scratch/mlevy/tests/aux_pop/cleanup_maint-5.6_merge-cice5_20191106-sync_with_cesm2_1_x_rel.intel

For the aux_pop_MARBL test lists on both machines, I just verified that all cases build and run [I didn't bother with making baselines].

User interface (namelist or namelist defaults) changes?

Quoting from #14

change sfwf_filename back to former value for CORE forcing rely on ROF_GRID for setting ebm_param_filename instead of datm_mode use appropriate transient forcing (listed above) in OMIP compsets restore ecosys related settings and nml vars in C1850ECO_ECOCESM20 compsets new nml vars fe_bioavail_frac_offset and dust_ratio_to_fe_bioavail_frac_r different for ocn_coupling=’partially-coupled’ than ocn_coupling=’full’ use o2_consumption_scalef_opt=’file_time_invariant’ only for ocn_coupling=’full’

---

Additional note: I originally tried to rebase cesm2_1_x_rel onto a fresh branch off master, but somehow ended up duplicating several dozen commits... which I didn't realize until I tried to put together a PR and saw it wouldn't merge into master cleanly. So I created a second fresh branch off master and resolved conflicts by copying the version of files from my original branch. So b810dee replaces the last five commits from that other branch and I've verified that the two branches themselves are identical. Further, all the testing was re-run using the branch that is actually in the PR.

[mnlevy1981]

The corresponding CIME PR is https://github.com/ESMCI/cime/pull/3462 and I realized I didn't explicitly mention the inputdata files I converted from netcdf-4 to cdf5:

1. ecosys_jan_IC_omip_POP_gx1v7_c190615.nc -> ecosys_jan_IC_omip_POP_gx1v7_c200323.nc in ocn/pop/gx1v6/ic/
2. ndep_ocn_omip_w_nhx_emis_gx1v6_1637-2019_c190522.nc -> ndep_ocn_omip_w_nhx_emis_gx1v6_1637-2019_c200324.nc in ocn/pop/gx1v6/forcing/
3. ndep_ocn_omip_w_nhx_emis_gx3v7_1637-2019_c190522.nc -> ndep_ocn_omip_w_nhx_emis_gx3v7_1637-2019_c200324.nc in ocn/pop/gx3v7/forcing/
4. ecosys_jan_IC_gx1v6_20190712.nc -> ecosys_jan_IC_gx1v6_20200325.nc in ocn/pop/gx1v6/ic/ (includes coccolithophore tracers)
5. ecosys_jan_IC_gx3v7_20190712.nc -> ecosys_jan_IC_gx3v7_20200325.nc in ocn/pop/gx3v7/ic/ (includes coccolithophore tracers)
6. ecosys_jan_IC_gx1v6_9p6z_20180124_3impCalc.nc -> ecosys_jan_IC_gx1v6_9p6z_20200325_3impCalc.nc in ocn/pop/gx1v6/ic/ (initial conditions for testing SPECTRA)
7. ecosys_jan_IC_gx3v7_9p6z_20180124_3impCalc.nc -> ecosys_jan_IC_gx3v7_9p6z_20200325_3impCalc.nc in ocn/pop/gx3v7/ic/ (initial conditions for testing SPECTRA)

I used my home-made netcdf comparison tool to verify that the data in the new files are identical to the old, and have added all the new files to the inputdata repository. I'd like to update cesm2_1_x_rel to use the newer versions of the first three files (a) for consistency between tags, and (b) as another sanity check that the files are bit-for-bit after the conversion; I'll work on that after this PR is merged.

[mnlevy1981]

@klindsay28 and I reviewed this and 86fb325 addresses the two minor issues raised in review. I'll merge it and tag it on Monday.