Support for torch.float weighted networks for FID and KID calculations.

[furkan-celik]

What does this PR do?

For evaluation of generated images in terms of how similar looking they are to the real images metrics FID is heavily used. However as pointed out in several papers, FID has some inherent flaws one of which is ImageNet InceptionV3 model is not suitable for all domains. Domains like medical MRI might be different enough to ImageNet that activation functions are not relevant to make FID comparison. For cases like this different domains use different variants of InceptionV3 model as feature extractor.

However, current implementation of FID and KID in torchmetrics only supports models with byteTensor weights as it is optimized for the tochmetrics' InceptionV3 model. In this PR I have given a bit more control to the developer so that it is possible to have better support for custom feature extractors with different dtypes and input sizes.

Details of this issue can be found in the following paper:
Liu, Shaohui, et al. "An improved evaluation framework for generative adversarial networks." arXiv preprint arXiv:1803.07474 (2018).

Fixes #<issue_number>

Before submitting

• [X I had issues with my implementations and didn't found a solution] Was this discussed/agreed via a Github issue? (no need for typos and docs improvements)
• [+] Did you read the contributor guideline, Pull Request section?
• [+] Did you make sure to update the docs?
• [+] Did you write any new necessary tests?

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

---

📚 Documentation preview 📚: https://torchmetrics--2483.org.readthedocs.build/en/2483/

[codecov]

Codecov Report

Merging #2483 (51dcc91) into master (5259c22) will decrease coverage by 31%.
The diff coverage is 0%.

Additional details and impacted files

@@           Coverage Diff            @@
##           master   #2483     +/-   ##
========================================
- Coverage      69%     38%    -31%     
========================================
  Files         307     307             
  Lines       17404   17410      +6     
========================================
- Hits        11989    6662   -5327     
- Misses       5415   10748   +5333     

[SkafteNicki]

Looking good :)