Update metrics to support both numpy and torch tensor inputs

[tijyojwad]

Convert inputs and inputs to torch tensors or numpy arrays before the call function based on whether the function uses scikit learn or torch functions for metric calculation

[ntadimeti]

@tijyojwad Is this still relevant? could you comment on what/how this PR is going to help with? Also, relevant files that need to be looked at for this change, if it needs to be addressed.

[tijyojwad]

This was intended for the metrics calculation classes. From what I recall there's a conditional in every metrics class that checks for whether or not the input array is numpy or a torch tensor. Python decorators are a neater way to do it so each of the classes can just call the decorator that does that check and conversion.

I think it's still relevant, but low priority.

[ntadimeti]

Gotcha! Makes sense.