Add convenience function to apply Hausfather et al. 2022 'screening' methodology

[jbusecke]

I just implemented the suggested screening methods of Hausfather et al. 2022 for an analysis of Sea Surface Trends in CMIP6.

They make a compelling argument to use this method widely when conducting CMIP-multi-model studies, and so maybe we ought to make this easy for the user? I am still thinking of what would be a good API here, but just wanted to note this here.

The base data is a mapping of source_id to bool provided here, one could filter datasets out on the datasets_dict/datatree level or provide a general version of what I did above, where instead of filtering we add new coodinates to each datasets that contain the necessary information for screening derived products later.