Add WeightLens input-invariant feature analysis

[Xmaster6y]

Summary

• add model-agnostic encoder-to-embedding, decoder-to-logit, and cross-layer feature projection APIs
• add deterministic WeightLens-compatible z-score outlier selection and optional batched forward validation
• support distinct input/output vocabularies, MPS-safe statistics, and no-grad analysis
• emit frozen, JSON-serializable analysis artifacts and export the public API from tdhook.weights
• cover projections, candidate pooling, validation, serialization, shape errors, and reference-formula agreement with toy-model tests

Validation

• uv run pre-commit run --all-files
• just tests (642 passed, 1 skipped; 98.36% total coverage; 100% coverage for weight_lens.py)

Closes #124

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.36%. Comparing base (8f22922) to head (7eb5287).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #130      +/-   ##
==========================================
+ Coverage   98.31%   98.36%   +0.05%     
==========================================
  Files          52       53       +1     
  Lines        3968     4091     +123     
==========================================
+ Hits         3901     4024     +123     
  Misses         67       67              

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

[cubic-dev-ai]

All reported issues were addressed across 3 files

_{You’re at about 96% of the monthly reviewed-line limit. You may want to disable incremental reviews to conserve quota. Reviews will continue until that limit is exceeded. If you need help avoiding interruptions, please contact contact@cubic.dev.}

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 95f2fac874

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-08-31T16:29:54.932798Z	95f2fac	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.