Relative training data - example notebook

[Peter9192]

This notebook illustrates what we'd like to do for #184. Three main steps:

• Create dummy data
• Store complex types in dataframe and then do feature extraction on those types
• Make duplicates of training data in a "sliding window" fashion, such that you have training data for different "snapshots" in the growing season.