Information imbalance for representation similarity

[Xmaster6y]

What does this PR do?

Key insights about the PR.

Linked Issues

• Linked to [Roadmap] Implement new methods #34

Checklist

• I have read the CONTRIBUTING guide.
• I have added tests for my changes if needed.
• I have updated the documentation if needed.

---

Summary by cubic

Adds InformationImbalanceEstimator for directional representation similarity using nearest-neighbor ranks, addressing #34. Exposes it in tdhook.latent, updates the representation similarity notebook with an Information Imbalance section, and adds tests.

• New Features
  • Introduces tdhook.latent.representation_similarity.InformationImbalanceEstimator computing A→B and B→A from nearest-neighbor rank preservation.
  • Supports (N, D) and batched (..., N, D), configurable p, dtype promotion, and input validation (matching batch shapes/devices, N≥2).
  • Exported via tdhook.latent and tdhook.latent.representation_similarity; notebook adds examples and API notes alongside CkaEstimator.
  • Tests cover custom keys, batching, determinism, dtype promotion, and error cases.

^{Written for commit a38cf63. Summary will update on new commits.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.39%. Comparing base (bc7978d) to head (a38cf63).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #49      +/-   ##
==========================================
+ Coverage   97.30%   97.39%   +0.08%     
==========================================
  Files          41       42       +1     
  Lines        2489     2572      +83     
==========================================
+ Hits         2422     2505      +83     
  Misses         67       67              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[cubic-dev-ai]

0 issues found across 1 file (changes from recent commits).

_{Requires human review: Introduces a new mathematical estimator (InformationImbalanceEstimator). New feature implementations of algorithms require human verification for correct domain logic.}

[cubic-dev-ai]

No issues found across 5 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Requires human review: This PR introduces a new feature (InformationImbalanceEstimator) with significant logic and a new public API, which warrants human review for implementation correctness and design.}