[x] separate figure for linear/nodal, makes figure code less complex
[x] how to handle all/valid distinction? >> Maybe just plot all yearly means after figure creating? Or separate figure also
[x] clip timeseries after deriving statistics so the desired dataframes are all in one dict (means for all years, but model fit only after physical break). Assert both mean timeseries and model fit on tstart an include ext
[x] define line colors per dictionary key in plotting function (e.g. all model fit is blue line+dots, all yearly means is red crosses, etc) >> cmap(0)
[x] add validation timeseries to both figures before saving >> only one figure