Address tabulation performance

[GregoryPetrochenkov-NOAA]

This pull request addresses #162 .

Changes

• Two-class, multi-class, and continuous notebooks: Updated backend provider for plots
• pyproject.toml: Updated version
• gval_xarray: Updated workflow for computing cross-tabulation tables from agreement maps, added functionality to parse string attributes if they exist and convert relevant attributes to strings if need be
• agreement.py: Added pairing dictionary to agreement map
• compute_categorical_metrics.py: get all unique values if they are not lists (issue to be addressed)
• tabulation.py: Remove xarray spatial crosstab and instead use flox groupby count
• spatial_alignment.py: Use odc reproject (generalizable for numpy and dask) instead of reproject match
• categorical_statistics.py: Handle all invalid input quietly
• schemas.py: Allow candidate and benchmark values to be null and account for future list (issue to be addressed)
• visualize.py: Change backend provider
• test compare/accessors/catalogs cases compare/accessors: Modified tests given change in workflow