PR #1 adds a wrapper for the thermaltime model and transforms the incoming data (sklearn compatible (X, y)) to the structure understood by pyphenology see https://pyphenology.readthedocs.io/en/master/data_structures.html#phenology-environmental-data. Then inside pyphenology we go back to sklearn data structure see temperature_only_data_prep. We should explore if we can skip the internal transformation for all models, especially M1 and Naive that use other predictors besides daily mean temerature.