Validate sample with ml configuration

[peterdudfield]

Detailed Description

In order to avoid seeing a matrix multiplication errors, it would be good to validate the sample going into a model.

Context

• we dont want training to slow down
• we dont want R&D to slow down

Possible Implementation

• add a function that can be called by the app, or other scripts.
• This function will take in a sample, and the ml configruation like this
• Check for nwp and satellite
• Check the time dimension is the right size for the model
• Check the channel dimension is the right size
• Check the width and height is the right size
• For site data, check there is enough data for the model. This might be linked with the time or sun features

[itsbharatj]

Hey @peterdudfield, I will start working on it right away.

I understood the issue; as this is my first time contributing to OCF, can you tell where this function would be implemented in the codebase?

Thank You

[peterdudfield]

It perhaps could be in its own file, under pvnet/validation.py