GeoCarb priors code requires update to support different CO sources

[joshua-laughner]

Describe the bug
Since the update to ginput to handle GEOS FP-IT or GEOS IT data, the geocarb sub-command will crash. This is because it currently cannot get information about which model provided the CO in the met resampler file.

To Reproduce
Assuming all the input files are in the current directory, and the resampler file is the standard testing one with data between 00:33Z and 01:23Z on 1 Jul 2015:

run_ginput.py geocarb \
  GEOS.fpit.asm.inst3_3d_asm_Nv.GEOS5124.20150701_0000.V01.nc4,GEOS.fpit.asm.inst3_3d_asm_Nv.GEOS5124.20150701_0300.V01.nc4 \
  ./geocarb_mock_resampler.h5 \
  ./geocarb-priors.h5 \
  --nprocs 12 \
  --no-truncate-mlo-smo \
  --fo2-file o2_dmf_data.dat \
  -vv

Error message

Traceback (most recent call last):
  File "/home/jlaugh/Notebooks/TCCON-research-notebooks/ginput-development/src/ginput/ginput/priors/acos_interface.py", line 430, in _prior_helper
    priors_dict, priors_units, priors_constants = tccon_priors.generate_single_tccon_prior(
  File "/home/jlaugh/Notebooks/TCCON-research-notebooks/ginput-development/src/ginput/ginput/priors/tccon_priors.py", line 2992, in generate_single_tccon_prior
    _, ancillary_trop = concentration_record.add_trop_prior(gas_prof, obs_utc_date, obs_lat, mod_file_data,
  File "/home/jlaugh/Notebooks/TCCON-research-notebooks/ginput-development/src/ginput/ginput/priors/tccon_priors.py", line 2110, in add_trop_prior
    prof_gas[:] = co * self.compute_co_scale(prof_pres=pres, prof_theta=theta,
  File "/home/jlaugh/Notebooks/TCCON-research-notebooks/ginput-development/src/ginput/ginput/priors/tccon_priors.py", line 2061, in compute_co_scale
    raise ValueError(f'Unknown CO source: {co_source}')
ValueError: Unknown CO source: GeosSource.UNKNOWN

[joshua-laughner]

Some options for how to handle this:

1. Assume that the input GEOS files are from the same model version that provided the CO.
2. Require a new attribute or variable be added to the met resampler file that indicates which version of GEOS was ingested to get the CO profiles.
3. Have the GEOS version used for the CO priors be a command line input.

I prefer 2, since that ensures that we always use the correct scaling for the CO provided. 3 would be acceptable, but not ideal. 1 is probably not a good idea, because we could easily take CO from GEOS IT but met from GEOS FP-IT, for example.

However, this will not be a priority until/unless GeoCarb is slated to fly, or a similar instrument has a use for that part of the code.