New metric: Perceptual Path Length

[SkafteNicki]

What does this PR do?

Fixes #1933

Before submitting

• 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--1939.org.readthedocs.build/en/1939/

[codecov]

Codecov Report

Merging #1939 (e07de45) into master (5d977e9) will decrease coverage by 44%.
The diff coverage is 71%.

Additional details and impacted files

@@           Coverage Diff            @@
##           master   #1939     +/-   ##
========================================
- Coverage      87%     43%    -44%     
========================================
  Files         266     268      +2     
  Lines       15280   15427    +147     
========================================
- Hits        13224    6603   -6621     
- Misses       2056    8824   +6768     

[SkafteNicki]

@Borda, @justusschock please re-review. After a second look at this, we already had a lot of the needed functionality already, so torch-fidelity is no longer needed for the implementation (only testing). A bit of code needed to be copied, but I have linked where needed to original implementation.

[Borda]

After a second look at this, we already had a lot of the needed functionality already, so torch-fidelity is no longer needed for the implementation (only testing). A bit of code needed to be copied, but I have linked where needed to original implementation.

I am fine with the code, my only concern is the dependency, can we somehow use the released package? what we really need form the master branch?

[SkafteNicki]

After a second look at this, we already had a lot of the needed functionality already, so torch-fidelity is no longer needed for the implementation (only testing). A bit of code needed to be copied, but I have linked where needed to original implementation.

I am fine with the code, my only concern is the dependency, can we somehow use the released package? what we really need form the master branch?

It's no longer a dependency of this metric, only torchvision is needed.

But we need it for testing to compare against. And we specific need master because the addition of PPL in torch-fidelity happened after latest release