Missing data in an observed flows will give a nan for the Nash-Sutcliffe

[Mayetea]

I was testing the calibration of models with the HYSETS_2020_ERA5 dataset and I found out that when a data is missing in the observed streamflow, the calibration fails and the Nash-Sutcliffe is set to nan. It gives back the calibrated parameters but I'm not sure they can be considered valid since the "fitness" is null. I don't know if this behavior is just when a flow is missing or when any data is missing, since the temperature and precipitation are all filled properly.

Finally I don't know if its Ouranos or a backend issue, so I'll open the issue here, but we can move it if its a backend issue.

Here's some image of the issue :

Streamflow missing :

Result :

[Mayetea]

If you want to reproduce the issue, the watershedId that I took was 23.

[huard]

The NSE is computed by RavenC itself, not the Python wrapper, so I suspect there isn't much we can do here, except raising the issue with James.