Regrid ECWMF data

[peterdudfield]

Simialr to #143, lets regrid the ECMWF data, so thats its the same as the training data

[Sukh-P]

On this: @devsjc am I right in thinking there was some interpolation done for ECMWF (NW India) archive data to go from 0.1 degrees to 0.05 degrees? Do you know if it was linear interpolation or another method? It would be good in this app to use the same method to try and reduce the difference between training and live