Tabulation performance

[GregoryPetrochenkov-NOAA]

As of now crosstab from xarray spatial is too slow and the workflow in incurring extra memory overhead. The new workflow should create the crosstab dataframe using flox groupby on the agreement map which contains the pairing dictionary as an attribute.

According to tests this should boost the speed up by more than 40%.