Stop giving the ML model satellite images of the _future_!

[JackKelly]

To test things, we showed the ML model satellite images of the future (i.e we showed the ML model satellite images after t0). We should probably stop doing that now that the code is working pretty well :)

[peterdudfield]

I had run an experiment for this

https://github.com/openclimatefix/predict_pv_yield/blob/issue/85-perciever-nwp/experiments/2021-10/2021-10-01/experiment.txt

Error went from 3% to 3.6%, bare in mind baseline is 5%, as a a rough idea

[JackKelly]

Awesome - thanks loads for running this experiment! Sounds like good news to me!