Refactor RC IFs to R's psi convention, fix HonestDiD VCV subsetting

Restructure _outcome_regression_rc, _ipw_estimation_rc, _doubly_robust_rc
to compute leading IF terms in R's unnormalized psi convention (using
mean_w_* = sum_w_*/n_all normalizers matching R's mean()), then convert
to library phi = psi/n_all at the boundary. Makes DRDID correspondence
explicit with R variable name comments.

Fix HonestDiD event_study_vcov subsetting: when filtering NaN-SE event
times, subset the VCV matrix to match the surviving rel_times (was using
the full unfiltered matrix, causing dimension mismatch on interior drops).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: igerber

diff_diff/honest_did.py

@@ -664,8 +664,16 @@ def _extract_event_study_params(
                 # Use full event-study VCV if available (Phase 7d),
                 # otherwise fall back to diagonal from SEs
                 if hasattr(results, "event_study_vcov") and results.event_study_vcov is not None:
-                    # event_study_vcov is indexed by sorted rel_times
-                    sigma = results.event_study_vcov
+                    vcov = results.event_study_vcov
+                    # VCV is indexed by ALL event times from aggregation;
+                    # rel_times may be a filtered subset (NaN-SE times dropped).
+                    # Subset VCV to match the surviving rel_times.
+                    all_event_times = sorted(results.event_study_effects.keys())
+                    if vcov.shape[0] == len(all_event_times) and len(rel_times) < len(all_event_times):
+                        idx = [all_event_times.index(t) for t in rel_times]
+                        sigma = vcov[np.ix_(idx, idx)]
+                    else:
+                        sigma = vcov
                 else:
                     sigma = np.diag(np.array(ses) ** 2)