Enable caching for distance calculation

[henrymartin1]

Calculating distances on a large dataset is an expensive operation, especially if we need to use the function calculate_haversine_length. This function is used by several functions (calculate_modal_split, predict_transport_mode) and additionally a user might want to have access to the length as well.

It would be great if we could implement a way to cache the calculation e.g., in a column of the input dataframe. I am not sure yet what would be a good concept that is still easy to use.
An idea would be to return the column if calculated as optional output if caching is activated.