#to do

in dowload.py:
- add target period to file name for forecast data - check
- catch errors in download when data are not availble or iridl is down
- fix CanSips download error
- implement stop button

- needs to be revised - should download data in native format, because then manually downloaded files could be used too



in forecast.py
- divergent scale for probability maps - centered on 0.33
- save tercile and cem categories as geojson
- include skill-masked plots


- make sure "native" IRI files and ECMWF files read properly


- save model configuration to output directory

- add "too dry to forecast" colour to plots
- limit CCA forecast for less than X locations/zones

- optimize skill calculation on gridded data - it takes much longer than calculations of the model


# wish list:
- implement download from CDS, and if not - implement download of CDS data from IRIDL - both require users to set up account
- multimodel forecast
- filling missing values in predictand
- regridding of gridded predictand to a coarser resolution if domain too large



