TSM Refactor stage 1: Numba equations, constants, and pyproject

[xaviernogueira]

Changes:

• Moved all TSM equations into their own numba JIT functions.
• Pulled constants out into typed dictionaries in their own module.
• Wrote a pyproject.toml file that works and allows the modules to be installed via pip install .
• Separated tests into a directory that should be passing, and a directory to store deprecated tests. All tests in /tests should be passing or in active development (currently empty).
• Updated the conda environment.yml file to include pytest and pyright, which is what I am using for static analysis.
• Made all arguments lower case for consistency and improved syntax highlighting (upper case assumes they are types not values).

Note that I implemented the model in another branch XRN_xarray_simlab. There I am exploring refactor design options (see below).

Next steps:

• @aufdenkampe (anyone can do this) pull out all equations used within other modules (NSM, GSM) into numba JIT functions following the pattern demonstrated in tsm/equations.py. I will loop back to this once figure out the architecture for xarray-simlab, but regardless we will need these functions JITed as I plan on using xarray.apply_ufunc(parallelize=True) where the numba functions are applied across xarray dataarrays.
• I need to finish implementing xarray-simlab. As I look closer, it is worth using the framework, as the idea I had to work with xarray without it is largely similar.
• Write tests for TSM. To leverage the existing answers I was thinking of making a 1 cell xarray and testing on that.