Add missing metrics

[sashavor]

As per @douwekiela's suggestion, we should find the blind spots that we have in terms of missing metrics, especially from domains like speech recognition and computer vision.

Suggestions are welcome below!

[sashavor]

We should probably look into GAN metrics as well, like Kernel Inception Distance (KID), Inception Score (IS) and Fréchet Inception Distance (FID) (maybe we should let people import them directly from a library like Torch Fidelity?

[sashavor]

How about RL metrics? e.g. https://analyticsindiamag.com/metrics-for-reinforcement-learning/

[sashavor]

Computer vision metrics: SSIM PSNR

There are various object detection metrics implemented by Tensorflow

[lvwerra]

Also listed a few libraries in #11 e.g. NetworkX for graph metrics.

[andreaferretti]

For image generation, also consider LPIPS as well as unpaired metrics such as NIQE, PIQE, Brisque, SR-Metric

[cceyda]

This is also an interesting method/library for evaluating text generation https://github.com/neulab/BARTScore