Update climatology functions following Xarray GroupBy enhancements

[kafitzgerald]

Deepak gave a great Pangeo Showcase the other week on updates to Xarray's GroupBy.

Slides and notebooks here: https://gist.github.com/dcherian/42411d8645573245045f0a55febf5fbe

These updates offer improved performance and increased flexibility for things like custom seasons (though don't address things like weighting).

It'd be good to go through our climatology functions to check for compatibility and/or whether or it may make sense to deprecate some of our functions now that upstream functionality has been enhanced.

GeoCAT Applications seems like a good place to begin exploring the latter bit (e.g. in NCAR/geocat-applications#150).

[kafitzgerald]

Ok, so in looking through our climatology related functions, I think the main thing we still offer over using Xarray directly is the temporal weighting and in some cases a bit less verbose syntax. This is probably enough to warrant keeping them around though.

We have fallen somewhat behind with respect to custom seasons, but could fairly easily support some additional options and/or implement custom seasons for additional comp functions leveraging the new Xarray functionality described here: https://earthmover.io/blog/ergonomic-seasonal-grouping-and-resampling-in-xarray

The other thing we could look into is adding some notes about performance / optimization for these functions (e.g. dask+flox should help quite a lot, but isn't a required dependency and probably isn't something we want to require).

I'm going to go ahead and add the Xarray examples (where they don't already exist) over in applications, but probably won't get to these custom seasons enhancements.