Bugfix on the production of diagnostic variables in rollout_to_netcdf.py + credit.postblock update + other minor bugfix

[yingkaisha]

New feature

• The water budget calculation is split from the GlobalMassFixer; it now has its own GlobalWaterFixer.
• GlobalMassFixer, GlobalWaterFixer, and GlobalEnergyFixer can now run outside of the model (i.e., they can run inside credit.trainer and rollout_to_netcdf.py).

Bugfix

• Diagnostic variables can now be produced properly in rollout_to_netcdf.py
• credit.output now produces files with coordinate names of latitude and longitude rather than 'lat/lon'
• credit.physcis_constants values were updated to align with CAM6
• A small bugfix GlobalMassFixer to handle "half level" vertical coordinates

Note

• The purpose of this PR is to resolve diagnostic-variable-related errors in rollout_to_netcdf.py. It maintains the structure of the current script and touches diagnostic variables only. rollout_to_netcdf.py can be improved fundamentally, but not in this PR.
• This PR can be verified as follows:

python applications/train.py -c config/example_physics_single.yml
python applications/rollout_to_netcdf.py -c config/example_physics_single.yml

(i.e., produce model weights --> use weights to verify rollout)

[yingkaisha]

You may see some crossformer.py commits by accident.

I have pushed new commits to align it with the main branch. There are no Wxformer updates in this PR

[yingkaisha]

Ready

[yingkaisha]

Merge conflicts resolved

[djgagne]

The new postblock physics and diagnostic updates look good to me. I fixed the CI issue and ran some formatting fixes and removed some unused imports and variables that were flagged by PyCharm. I also want to mention a couple of formatting/style conventions to encourage going forward

1. We use Google-style docstrings across MILES projects, which follow the following format:

"""
short description of what function/method does

Args
    first_arg (type): blah
    second_arg (type): blah
Returns
    variable (type): description of thing being returned
"""

Please use double-quotes instead of single quotes for consistency purposes. Both types are interchangeable in Python as long as one uses the same type at the beginning and end of a string, but I suspect style guides encourage double quotes because in some languages single quotes can only be used for character types.