Bump version to 2.8.4

[igerber]

Summary

• Bump version from 2.8.3 to 2.8.4 across all version locations
• Add CHANGELOG entry covering PRs Update ImputationDiD tutorial to show pretrends=True event study #258, Add SDR replicate method and FPC support for ImputationDiD/TwoStageDiD #259, Add bootstrap lonely_psu adjust, CV on estimates, weight trimming, and pretrends+survey #260, Fix survey tutorial warnings, narrative, and chart rendering #261:
  • SDR replicate method for ACS PUMS (Phase 8a)
  • FPC support for ImputationDiD/TwoStageDiD (Phase 8b)
  • Lonely PSU "adjust" in bootstrap (Phase 8d)
  • CV on estimates, weight trimming utility (Phase 8e)
  • ImputationDiD pretrends + survey (Phase 8e)
  • Survey tutorial and ImputationDiD tutorial updates

Methodology references (required if estimator / math changes)

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

Validation

• Tests added/updated: No test changes (version bump only)
• Backtest / simulation / notebook evidence: N/A

Security / privacy

• Confirm no secrets/PII in this PR: Yes

---

Generated with Claude Code

[github-actions]

Overall Assessment

✅ Looks good

Executive Summary

• This PR does not modify estimator, weighting, variance, or inference code; methodology review is limited to validating the new release-note text.
• The methodology-bearing changelog items align with the existing Methodology Registry and code docstrings for SDR, survey-aware pretrends, FPC support, and bootstrap lonely_psu="adjust".
• The version bump is internally consistent across the Python package, Rust crate, project metadata, and LLM docs surfaces.
• I found one documentation consistency issue: the new 2.8.4 changelog says several survey features shipped, but the survey tutorial and survey roadmap still describe those same items as unsupported or pending.
• No P0/P1 correctness, methodology, or security findings in the diff.

Methodology

• No findings. The only methodology-bearing change is the release-note block in CHANGELOG.md#L10. The affected methods it names already match the registry and code documentation for survey-aware pretrends in docs/methodology/REGISTRY.md#L853 and diff_diff/imputation.py#L84, FPC support in docs/methodology/REGISTRY.md#L884 and docs/methodology/REGISTRY.md#L963, bootstrap lonely_psu="adjust" in docs/methodology/REGISTRY.md#L2218 and diff_diff/bootstrap_utils.py#L430, and SDR in docs/methodology/REGISTRY.md#L2263 and diff_diff/survey.py#L86.

Code Quality

• No findings. The version bump is consistent across pyproject.toml#L7, diff_diff/init.py#L214, rust/Cargo.toml#L3, and docs/llms-full.txt#L5. I did not find any remaining 2.8.3 version strings.

Performance

• No findings. No runtime code paths or dependency behavior changed.

Maintainability

• No findings beyond the documentation consistency item noted below.

Tech Debt

• No findings. This PR does not introduce new deferred work or untracked correctness debt.

Security

• No findings. The diff contains no secrets, permission changes, or security-sensitive behavior changes.

Documentation/Tests

• Severity: P2. Impact: the new release notes say 2.8.4 includes FPC support for ImputationDiD/TwoStageDiD, survey-aware ImputationDiD pretrends, and bootstrap lonely_psu="adjust", but the survey tutorial still marks ImputationDiD and TwoStageDiD as “Partial (no FPC)” and the survey roadmap still describes 8b/8d/8e as pending work. That leaves the release documentation internally contradictory. References: CHANGELOG.md#L10, docs/tutorials/16_survey_did.ipynb#L530, docs/survey-roadmap.md#L229, docs/survey-roadmap.md#L251, docs/survey-roadmap.md#L263. Concrete fix: update the survey compatibility matrix and revise or mark those roadmap items as completed in 2.8.4 so the docs agree on current feature status.
• No test-blocking finding. For a version-bump-only PR, the absence of new code tests is acceptable.