Bacon empty-comparisons guard, document zero-weight df, complete test coverage

- Bacon: raise ValueError when all 2x2 comparisons dropped after
  zero-weight filtering (prevents empty decomposition results)
- Document survey df convention: total n (including zero-weight rows)
  matches R's survey::degf() after subset() — preserves design structure
- Add SyntheticDiD and TROP replicate rejection tests
- Add CS ipw/dr (no covariates) replicate positive-path tests

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

Author: igerber

diff_diff/bacon.py

@@ -585,6 +585,13 @@ def fit(
                 weights=survey_weights,
             )
 
+        if not comparisons:
+            raise ValueError(
+                "No valid 2x2 comparisons remain after filtering. "
+                "All cells have zero effective weight or insufficient data. "
+                "Check subpopulation/domain definition."
+            )
+
         # Normalize weights to sum to 1
         total_weight = sum(c.weight for c in comparisons)
         if total_weight > 0:

docs/methodology/REGISTRY.md

@@ -2093,6 +2093,11 @@ Domain estimation preserving full design structure.
 - **Note:** Weight validation relaxed from "strictly positive" to
   "non-negative" to support zero-weight observations. Negative weights
   still rejected. All-zero weight vectors rejected at solver level.
+- **Note:** Survey df and variance adjustment factors use total n
+  (including zero-weight rows), matching R's `survey::degf()` convention
+  after `subset()`. This preserves the design structure for correct
+  variance estimation. Zero-weight rows contribute zero-valued scores
+  to the sandwich meat but are counted in df = n_PSU - n_strata.
 - **Note:** For replicate-weight designs, `subpopulation()` zeros out both
   full-sample and replicate weight columns for excluded observations,
   preserving all replicate metadata.

tests/test_survey_phase6.py

@@ -1458,6 +1458,57 @@ def test_bacon_replicate_rejected(self):
                 first_treat="first_treat", survey_design=sd,
             )
 
+    def test_synthetic_did_replicate_rejected(self):
+        """SyntheticDiD rejects replicate-weight designs."""
+        from diff_diff.synthetic_did import SyntheticDiD
+        data, sd = self._replicate_sd_and_data()
+        data["treated"] = (data["first_treat"] > 0).astype(int)
+        with pytest.raises(NotImplementedError):
+            SyntheticDiD().fit(
+                data, outcome="outcome", unit="unit", time="time",
+                treatment="treated", survey_design=sd,
+            )
+
+    @pytest.mark.slow
+    def test_trop_replicate_rejected(self):
+        """TROP rejects replicate-weight designs."""
+        from diff_diff.trop import TROP
+        data, sd = self._replicate_sd_and_data()
+        data["treated"] = (data["first_treat"] > 0).astype(int)
+        with pytest.raises(NotImplementedError):
+            TROP().fit(
+                data, outcome="outcome", unit="unit", time="time",
+                treatment="treated", survey_design=sd,
+            )
+
+
+class TestCSReplicateMethodCoverage:
+    """Test that CS replicate weights work for all supported estimation methods."""
+
+    @staticmethod
+    def _make_cs_replicate_data():
+        data, rep_cols = TestEstimatorReplicateWeights._make_staggered_replicate_data()
+        sd = SurveyDesign(
+            weights="weight", replicate_weights=rep_cols,
+            replicate_method="JK1",
+        )
+        return data, sd
+
+    @pytest.mark.parametrize("method", ["ipw", "dr"])
+    def test_cs_replicate_ipw_dr_no_covariates(self, method):
+        """CS replicate weights work for ipw/dr without covariates."""
+        from diff_diff import CallawaySantAnna
+        data, sd = self._make_cs_replicate_data()
+        result = CallawaySantAnna(
+            estimation_method=method, n_bootstrap=0,
+        ).fit(
+            data, "outcome", "unit", "time", "first_treat",
+            survey_design=sd,
+        )
+        assert np.isfinite(result.overall_att)
+        assert np.isfinite(result.overall_se)
+        assert result.survey_metadata is not None
+
 
 # =============================================================================
 # Effective-sample and d.f. consistency tests