Refactor bundled example dataset interface

[dylanhmorris]

Right now, each dataset has its own user-exposed loader function. I think it would be better to follow the pattern ArviZ takes: a single user-exposed az.load_arviz_dataset() that takes the name of the dataset to load as a string, plus an az.list_datasets() method that enumerates the available datasets, with description and documentation.

Thoughts @damonbayer @cdc-mitzimorris?

[mitzimorris]

az.list_datasets() method that enumerates the available datasets, with description and documentation.

what organization do we assume for the datasets? where does the dataset documentation come from?

currently, the datasets in pyrenew/datasets are organized into directories, and the hospital and wastewater data directories contain a README.md file. PR #758 adds a script to pyrenew/datasets - datagen_he_CA_120.py which creates a new directory synthetic_CA_120. what would this script need to generate so that it would work for this proposal?