Bump version to 2.8.3, add Phase 8 survey maturity roadmap

[igerber]

Summary

• Bump version from 2.8.2 to 2.8.3 across all version locations
• Add CHANGELOG entry covering PRs Add pre-period event study coefficients to ImputationDiD and TwoStageDiD #252, Add replicate weight support to 7 estimators #253, Add UserWarning emissions for 8 silent operations #256:
  • Silent operation warnings (8 items — TROP fallback, TwoStage NaN masking, CS pair skipping, etc.)
  • Replicate weight expansion to 7 estimators (4/13 → 11/13 coverage)
  • ImputationDiD/TwoStageDiD pre-period event study coefficients
• Add Phase 8 survey maturity roadmap to docs/survey-roadmap.md:
  • 8a: SDR replicate method (ACS PUMS)
  • 8b: FPC in ImputationDiD/TwoStageDiD
  • 8c: Silent operation warnings (done in this release)
  • 8d: Lonely PSU "adjust" in bootstrap
  • 8e: Survey diagnostics (CV, weight trimming, pretrends+survey)
  • 8f: Survey compatibility matrix
  • 8g: Documentation-only items

Methodology references (required if estimator / math changes)

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

Validation

• Tests added/updated: No test changes (version bump + roadmap)
• 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

• No estimator implementation, weighting math, variance/SE code, defaults, or assumption checks changed in this diff.
• Version propagation is consistent across the package, Rust crate, and machine-readable docs at diff_diff/init.py#L213, pyproject.toml#L7, rust/Cargo.toml#L3, and docs/llms-full.txt#L5.
• P2: the new Phase 8b roadmap text misdescribes the current survey-variance path for ImputationDiD and TwoStageDiD as existing TSL support, which does not match the Methodology Registry or estimator docstrings.
• P2: after adding the 2.8.3 release notes, the survey roadmap still says replicate weights are rejected for seven estimators that the changelog, Methodology Registry, and test suite show are supported.
• P3: Phase 8c is added as planned work even though the same release notes and current code show those warnings are already implemented.

Methodology

• Severity: P2. Impact: The newly added Phase 8b text in docs/survey-roadmap.md#L229 lines 229-236 says ImputationDiD and TwoStageDiD already have a TSL variance path and frames FPC as simply threading fpc through it. That does not match the canonical methodology notes, which say these estimators currently use survey-weighted Theorem 3 / GMM variance with PSU as the cluster variable, while strata only affects survey degrees of freedom; full stratified TSL/FPC support is not implemented yet in either estimator. See docs/methodology/REGISTRY.md#L884, docs/methodology/REGISTRY.md#L963, diff_diff/imputation.py#L213, and diff_diff/two_stage.py#L209. Concrete fix: rewrite 8b to describe the current implementation accurately and avoid implying that FPC support is just parameter plumbing through an existing TSL path.

Code Quality

• Severity: None. Impact: The actual version bump is internally consistent across all touched version surfaces. Concrete fix: None.

Performance

• Severity: None. Impact: This is a documentation/version diff; no runtime paths changed. Concrete fix: None.

Maintainability

• Severity: P3. Impact: Survey-support status is duplicated across multiple docs, and this PR updates only some of them. The drift is already visible between docs/survey-roadmap.md#L93 lines 93-108 and the canonical support matrix in docs/methodology/REGISTRY.md#L2300. Concrete fix: keep the detailed support matrix in one canonical source and have roadmap/tutorial docs link to it rather than restating it.

Tech Debt

• Severity: None. Impact: No new untracked technical debt in code paths; the only new debt is documentation drift noted above. Concrete fix: None.

Security

• Severity: None. Impact: No secrets, dependency changes, or security-relevant code paths are introduced by this diff. Concrete fix: None.

Documentation/Tests

• Severity: P2. Impact: The roadmap still states that replicate weights are rejected for DifferenceInDifferences, TwoWayFixedEffects, MultiPeriodDiD, StackedDiD, SunAbraham, ImputationDiD, and TwoStageDiD at docs/survey-roadmap.md#L93 lines 93-108, but the new release notes say those seven estimators were added in 2.8.3 at CHANGELOG.md#L22 lines 22-24, the Methodology Registry lists them as supported at docs/methodology/REGISTRY.md#L2300, and the test suite exercises them in tests/test_replicate_weight_expansion.py#L1. Concrete fix: update the Phase 6 replicate-weight limitations block to match the registry/tests, or remove the duplicated matrix and point readers to the registry.
• Severity: P3. Impact: Phase 8c is presented as planned future work in docs/survey-roadmap.md#L238 lines 238-249, but the same release notes mark those warnings as already shipped in CHANGELOG.md#L12 lines 12-21, and representative warning paths already exist in diff_diff/survey.py#L189, diff_diff/two_stage.py#L298, diff_diff/staggered.py#L1814, and diff_diff/trop_global.py#L372. Concrete fix: mark 8c as completed in 2.8.3 or move it out of the planned-work section.