Add function to_group_dataframe() to darts/timeseries.py that takes a list of TimeSeries and generates a "long" DataFrame (with the desired "backend") with static covariates (optional with default True, allow column select) and metadata (optional with default False, allow column select) added as columns.
For reference, the output should allow to be used by TimeSeries.from_group_df() to re-create the same TimeSeries.