Add MCS, adapt creation of L1 and L2 datacubes

[pat-schmitt]

Sorry for the big PR, but I needed to change many things in order to have everything ready for the validation framework.

The main changes are:

• L1 datacubes generation start now from OGGM gridded data, and their is a new method BindingsOggmModel.add_data_to_l1_datacube which adds different observations to the L1 datacubes. Current options:
  • Cryosat: from dtcg.datacube.cryotempo_eolis import DatacubeCryotempoEolis and call BindingsOggmModel.add_data_to_l1_datacube(DatacubeCryotempoEolis)
  • WGMS: from dtcg.datacube.wgms import DatacubeWGMS and call dtcg_oggm.add_data_to_l1_datacube(DatacubeWGMS)
  • Surface Type: from dtcg.datacube.surface_type import DatacubeSurfaceType and call dtcg_oggm.add_data_to_l1_datacube(DatacubeSurfaceType)
  • I have not spent much time to deal with different time coordinates so at the moment the convention is t for Cryosat, t_wgms for WGMS and t_sfc_type
• L2 datacube generation now includes Monte Carlo Simulation (MCS) for uncertainty propagation. Their are three L2 datacubes available:
  • 'monthly': monthly volume, area, length and mass-balance
  • 'annual_hydro': annual volume, area, length and mass-balance, and annual and monthly runoff components
  • 'daily_smb': daily mass-balance
• GeoZarr export:
  • now metadata_mapping includes entries for all model output (data and coordinates) and for the missing observations of the OGGM gridded data. However, for some I added just TODO for now as I need to move forward. I will open an issue about this.
  • I adapted the add_layer function to the new output format of the L2 datacubes. Their is now one node just for the name (e.g. 'L2_Hugonnet_daily') with up to three leaves corresponding to 'monthly', 'annual_hydro' or 'daily_smb'

I will start now with the implementation of the actual validation framework, building on the changes in this PR.

[gampnico]

@pat-schmitt I will close this since all these changes are part of #47, unless you plan on making more changes here?

[pat-schmitt]

yes this can be closed, as everything is in the other PR