Fix CI review R3: add l_vec to compute_honest_did, end-to-end tests

- Add l_vec parameter to compute_honest_did() so the advertised
  custom-target path actually works (was missing from wrapper)
- Add test_honest_did_custom_l_vec_on_impact: l_vec=[1,0] targets
  on-impact effect, asserts original_estimate matches DID_1
- Add test_honest_did_with_trends_nonparam: end-to-end
  trends_nonparam + honest_did=True integration

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

Author: igerber

diff_diff/honest_did.py

@@ -2669,6 +2669,7 @@ def compute_honest_did(
     method: str = "relative_magnitude",
     M: float = 1.0,
     alpha: float = 0.05,
+    l_vec: Optional[np.ndarray] = None,
 ) -> HonestDiDResults:
     """
     Convenience function for computing Honest DiD bounds.
@@ -2683,6 +2684,12 @@ def compute_honest_did(
         Restriction parameter.
     alpha : float
         Significance level.
+    l_vec : np.ndarray, optional
+        Weight vector defining the scalar target ``theta = l_vec' tau``
+        over post-treatment horizons. Length must equal the number of
+        post-treatment periods. ``None`` (default) uses equal weights
+        (uniform average). To target the on-impact effect only (R's
+        default), pass ``np.array([1, 0, ..., 0])``.
 
     Returns
     -------
@@ -2694,7 +2701,7 @@ def compute_honest_did(
     >>> bounds = compute_honest_did(event_study_results, method='relative_magnitude', M=1.0)
     >>> print(f"Robust CI: [{bounds.ci_lb:.3f}, {bounds.ci_ub:.3f}]")
     """
-    honest = HonestDiD(method=method, M=M, alpha=alpha)
+    honest = HonestDiD(method=method, M=M, alpha=alpha, l_vec=l_vec)
     return honest.fit(results)
 
 

tests/test_chaisemartin_dhaultfoeuille.py

@@ -3385,6 +3385,49 @@ def test_honest_did_original_estimate_is_post_average(self):
         avg = np.mean([es[h]["effect"] for h in sorted(es.keys())])
         np.testing.assert_allclose(hd.original_estimate, avg, rtol=1e-10)
 
+    def test_honest_did_custom_l_vec_on_impact(self):
+        """compute_honest_did with l_vec=[1,0] targets on-impact effect."""
+        from diff_diff.honest_did import compute_honest_did
+
+        df = self._make_data()
+        with warnings.catch_warnings():
+            warnings.simplefilter("ignore")
+            r = ChaisemartinDHaultfoeuille(seed=1).fit(
+                df, "outcome", "group", "period", "treatment",
+                L_max=2,
+            )
+        # l_vec=[1, 0] targets only DID_1 (on-impact, R's default)
+        bounds = compute_honest_did(r, l_vec=np.array([1.0, 0.0]))
+        np.testing.assert_allclose(
+            bounds.original_estimate,
+            r.event_study_effects[1]["effect"],
+            rtol=1e-10,
+        )
+
+    def test_honest_did_with_trends_nonparam(self):
+        """End-to-end trends_nonparam + honest_did=True."""
+        rng = np.random.RandomState(42)
+        rows = []
+        for g in range(40):
+            state = g % 4
+            switches = g < 20
+            for t in range(7):
+                d = 1 if (switches and t >= 3) else 0
+                y = 10 + 2.0 * t + 5.0 * d + rng.normal(0, 0.5)
+                rows.append({
+                    "group": g, "period": t, "treatment": d,
+                    "outcome": y, "state": state,
+                })
+        df = pd.DataFrame(rows)
+        with warnings.catch_warnings():
+            warnings.simplefilter("ignore")
+            r = ChaisemartinDHaultfoeuille(seed=1).fit(
+                df, "outcome", "group", "period", "treatment",
+                L_max=2, trends_nonparam="state", honest_did=True,
+            )
+        assert r.honest_did_results is not None
+        assert np.isfinite(r.honest_did_results.ci_lb)
+
 
 # =============================================================================
 # Summary Phase 3 Rendering