...following yesterday's ctsm python organization meeting with
Bill, Erik, and Negin.

...according to mtgs with Bill, Erik, and Negin.
Here I moved the ModifyFsurdat class out of fsurdat_modifier.py into
its own file and placed both those files in a new directory:
/python/ctsm/modify_fsurdat

Also replacing default fsurdat_in and fsurdat_out to work with this test:
SMS_D_Ld1.f10_f10_mg37.I1850Clm50Sp.cheyenne_intel.clm-collapse_pfts_78_to_16_decStart_f10
although ultimately we plan to remove these defaults.

This allows potentially overriding variables that may get set by the
first function, which identifies a user-defined idealized land swath.

I also completed the list of fsurdat variables that I believe need
updating in that first function (def land_swath).

I still need to address various TODOs.

Change ceta from 450 to 358 for CTSM5.1 See Leo VanKamenhout's work on this in
ESCOMP#250. Note, this is used in the PPE work.
See van Kampenhout et al. (2017).  Turn Medlyn on for PHS off for clm51 and clm50.
Make max CO2 partial pressure consistent in Photosynthesis.
Change surface datasets to only use 78PFT versions
Fix GSSUN and GSSHA history variables so now not just writing missing values.
Changes FATES to run with use_nitrif_dentrif=T by default. Change dynlakes test
to methane/use_nitrif_=T.

Reordered function calls again to what makes sense to me right now

Testing with
SMS_D_Ld1.f10_f10_mg37.I1850Clm50Sp.cheyenne_intel.clm-collapse_pfts_78_to_16_decStart_f10

For explanation see github post after this commit

Work in progress. Nevertheless, the code runs and the CTSM test
SMS_D_Ld1.f10_f10_mg37.I1850Clm50Sp.cheyenne_intel.clm-collapse_pfts_78_to_16_decStart_f10
passes.

A few other small modifications.
Python code runs and this test
SMS_D_Ld1.f10_f10_mg37.I1850Clm50Sp.cheyenne_intel.clm-collapse_pfts_78_to_16_decStart_f10
passes.

If True, then most fsurdat variables get modified, some with user-selected
values and some with default settings.
If False, then only user-selected variables change; other fsurdat variables
remain unchanged.

...because I placed it in utils.py in an earlier commit.

Still setting the 4D vars (lai, sai, and hgts) to 0 when
dom_nat_pft = 0 (bare ground).

Also now reading .cfg file's section name(s) on the fly.

As far as I can tell, this simplifies without loss of functionality.

Code didn't work correctly without the return statmt.
Removed leading underscores from _file and _not_rectangle
bc I use them outside of the class, ie they are not private.

Changes needed to get aux history output with cmeps

These changes are needed to create auxiliary land history files when the
new cmeps configuration variable is set to true. histaux_l2x1yrg =
.true.

The changes needed in CTSM enable lnd2glc fields to be sent to the
mediator even if glc is not present - based on querying the
histaux_l2x1yrg setting.

This includes a related change to set most lnd -> mediator fields to
spval over ocean, instead of sending zero values over ocean.

Currently gives errors:
- No such file or directory: 'rm ../tools/modify_fsurdat/test_modify.nc'
- No such file or directory: 'rm ../tools/modify_fsurdat/test_modify_minimal.nc'

Does not do much, yet, but runs successfully

sys test all not entirely done

Also add @staticmethod before def _get_not_rectangle.
These changes were pending before Bill's latest code review.

Remove _PLACEHOLDER in same.
Add relevant comment in get_config_value's docstring in utils.py.

Worked on this commit with Bill

Followed the approach used in
cime_config/SystemTests/fsurdatmodifyctsm.py that creates cfg files
starting from the pre-existing modify_template.cfg rather than requiring
separate pre-existing .cfg files for every test

In the process I updated function zero_nonveg to operate on the
rectangle (as all the other functions) rather than on the whole domain.
Doing so makes this function redundant if idealized = True, so I removed
it from the system test test_allInfo at this time.

Thanks to Brian Vanderwende in CISL for guidance on this. This is
still a bit of a kludge, and is hard-coded for cheyenne, but at least
it seems to work for now.

Run fsurdat_modifier via an appropriate python version on cheyenne

Gets Bill's mods to cime_config/SystemTests/fsurdatmodifyctsm.py
as recommended by Brian Vanderwende

Not all fsurdat files contain the variable F0, so I removed this
variable from the list of variables modified by fsurdat_modify

Store netCDF files via Git LFS

...plus corrected a couple of typos

FSURDATMODIFYCTSM_D_Mmpi-serial_Ld1.5x5_amazon.I2000Clm50SpRs.cheyenne_intel

...instead of TestFsurdatModifier

...from test_unit_modify_fsurdat.py and add to them is_list = False test.
Also break up lonRange0To360 tests to a test per case.

...or np.any(np.not_equal(arr1, arr2))