Generalize sklearn wrapper

[Peter9192]

#1 adds a wrapper for the thermaltime model. The core functionality consists in transforming the incoming data that's organized according to sklearn standards (X, y) to the format understood by pyphenology (see https://pyphenology.readthedocs.io/en/master/data_structures.html#phenology-environmental-data). It would be nice if we could generalize this translation so the wrapper can work with all pyphenology models. To this end, we need to come up with a standard convention for organizing/naming the input columns such that we know how to pivot it.