Release v2.9.0

[igerber]

Summary

• Bump version to 2.9.0 across all version locations (pyproject.toml, init.py, Cargo.toml, llms-full.txt)
• Add CHANGELOG entry for v2.9.0 covering: WooldridgeDiD (ETWFE), EfficientDiD survey+covariates, survey real-data validation, label-gated CI, documentation dependency map, and associated fixes

Methodology references

• N/A — version bump only, no methodology changes in this PR

Validation

• No test changes — release bookkeeping only

Security / privacy

• Confirm no secrets/PII in this PR: Yes

---

Generated with Claude Code

[github-actions]

Overall Assessment

✅ Looks good — no unmitigated P0/P1 findings. This PR is release bookkeeping plus changelog text; it does not modify estimator code, weighting, variance/SE computation, identification checks, or defaults.

Executive Summary

• The code changes are metadata-only: version bumps in pyproject.toml:7, diff_diff/init.py:217, rust/Cargo.toml:3, and docs/llms-full.txt:5.
• Release metadata is internally consistent, and docs versioning already flows from diff_diff.__version__ in docs/conf.py:19.
• The changelog’s cited subpopulation df deviation is explicitly documented in the Methodology Registry at docs/methodology/REGISTRY.md:2472.
• Minor documentation-only issues remain: the changelog slightly overstates how explicitly the TWFE survey-validation difference is documented, and the roadmap still says the current release is v2.8.4 at ROADMAP.md:11.

Methodology

• Severity: P3; Impact: CHANGELOG.md:19 says both survey-validation differences are “documented in REGISTRY.md,” but only the subpopulation df case is explicitly labeled there at docs/methodology/REGISTRY.md:2472. The TWFE item is only indirectly supported by the absorbed-FE methodology notes at docs/methodology/REGISTRY.md:240 and docs/methodology/REGISTRY.md:2273. This is documentation clarity only; no estimator or inference logic changes in this PR. Concrete fix: either add an explicit TWFE survey-validation note to REGISTRY.md or soften the changelog wording to say the differences are documented across the methodology docs.

Code Quality

• No findings. The touched version surfaces are synchronized at pyproject.toml:7, diff_diff/init.py:217, rust/Cargo.toml:3, docs/llms-full.txt:5, and the compare link is present at CHANGELOG.md:1155.

Performance

• No findings. No runtime or hot-path code changed.

Maintainability

• No findings. Version display in the built docs is already derived from package metadata via docs/conf.py:19, which reduces release-bump drift.

Tech Debt

• Severity: P3 (tracked in TODO.md); Impact: this PR touches docs/llms-full.txt, but .txt AI guides are still outside snippet CI coverage, which is already tracked at TODO.md:100. This is mitigated per the review policy. Concrete fix: none required for approval; the existing TODO item is sufficient tracking.

Security

• No findings. The diff is metadata/docs only and does not add secret-handling, dependency, network, or execution surface.

Documentation/Tests

• Severity: P3; Impact: the roadmap still presents v2.8.4 as the current status at ROADMAP.md:11, while this PR advances release metadata to 2.9.0. That creates minor user-facing version inconsistency across docs. Concrete fix: update the hard-coded version in ROADMAP.md, or remove the version number from that sentence so future release PRs do not need to keep it in sync.
• No test blocker. This PR does not change executable estimator behavior, so the absence of new methodology tests is acceptable for a release-bookkeeping PR.