Port R did package tests and fix not_yet_treated control group bugs

Port ~21 tests from R's `did` package (bcallaway11/did) to validate
CallawaySantAnna estimator. Tests organized as Tier 1 (Python DGP, loose
tolerance, always-run) and Tier 2 (R golden values, strict tolerance).

Found and fixed two bugs via the ported tests:
- not_yet_treated control group now works without never-treated units
  (requires ≥2 treatment cohorts)
- not_yet_treated control mask now uses max(t, base_period) to prevent
  contamination when base_period="universal" exceeds evaluation period

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

Author: igerber

TODO.md

@@ -154,6 +154,18 @@ Spurious RuntimeWarnings ("divide by zero", "overflow", "invalid value") are emi
 
 ---
 
+## Feature Gaps (from R `did` package comparison)
+
+Features in R's `did` package that block porting additional tests:
+
+| Feature | R tests blocked | Priority |
+|---------|----------------|----------|
+| Repeated cross-sections (`panel=FALSE`) | ~7 tests in test-att_gt.R + test-user_bug_fixes.R | Medium |
+| Sampling/population weights | 7 tests incl. all JEL replication | Medium |
+| Calendar time aggregation | 1 test in test-att_gt.R | Low |
+
+---
+
 ## Performance Optimizations
 
 Potential future optimizations: