Extend evaluation metrics beyond MAE

[DresdenGman]

Hey, I noticed the TODO in metrics.py about adding more metrics beyond MAE. I'd like to help with that.

I'm thinking of adding RMSE, mean bias error, and normalized versions of MAE/RMSE (divided by capacity) — all as standalone functions in the same file, no new dependencies. Would also rewrite the existing test to use actual assertions instead of the current smoke test.

Quick questions:

• Should these accept pandas Series as well, or numpy-only is fine?
• For empty inputs, raise ValueError or return NaN?

Happy to keep it small and focused. Let me know if this works for you.