modelbuilder improve ERA5 experience

[veenstrajelmer]

A non-wgs84 grid resulted in the following Delft3D FM error:

** INFO   : External Forcing or Initialising 'airpressure_windx_windy_charnock' from file 'era5_msl_u10n_v10n_chnk_20030301to20030331_ERA5.nc'.
** WARNING: Variable 'air_pressure' in NetCDF file 'era5_msl_u10n_v10n_chnk_20030301to20030331_ERA5.nc requires 'projection_x_coordinate' and 'projection_y_coordinate'.
** WARNING: Variable 'air_pressure' in NetCDF file 'era5_msl_u10n_v10n_chnk_20030301to20030331_ERA5.nc coordinates variable valid_time referenced but not found
** ERROR  : flow_initexternalforcings: Error while initializing quantity: airpressure_windx_windy_charnock . Check preceding log lines for details.

This was due to the fact that the software got a cartesian grid and expected a cartesian ERA5 netcdf, which was not the case. To avoid this, there are some steps to be done.

Todo:

• add comment to modelbuilder notebook: "only EPSG:4326 is supported by the CMEMS/ERA5 download functions"
• update docstring of dfmt.preprocess_ERA5() to state current relevance (expver dim and scaling/offset not present anymore, but was for some intermediate files) (valid_time renaming necessary to be consistent with old downloads)
• request valid-time rename to time at ecmwf forum >> never mind
• warning if era5 expver dimension is found? Or if scaling/offset is found? >> not needed, will probably be handled fine and preprocess_ERA5 is now optional.
• add safeguard to `dfmt.download_ERA5() for case with date_min>date_max
• rename valid_time in coordinates, this avoids confusion when supplying cartesian network >> no, too much effort compared to the benefit
• notebook: add lon-min etc input arguments to function (now order of passing matters)