Is your feature request related to a problem? Please describe.

When comparing scores with Quaver, there is a mismatch in the definition of forecast steps for the WeatherGenerator and the ECMWF convention.

Forecast step WG range from 1,2,3,4,5
According to ECMWF's convention the step should be the time distance from the initialization time, like e.g. 6, 12, 18, 24 etc..

This convention would be really convenient when comparing runs produced with different len_hrs.

Describe the solution you'd like

The safest place to handle this would be the ZarrIO class. The preferred solution would be to modify the forecast_step definition in the barrio class and multiply it by len_hrs.

Describe alternatives you've considered

One solution would be to add it to the evaluation package, but then there's a mismatch between the forecast_steps in the plot_config and the ones visible while browsing the ZarrIO, which would add some confusion

Additional context

I can implement it, but I need some hints from @grassesi

Organisation

No response