Fix CI review Round 13: per-period path uses raw Y under trends_linear

P1: When trends_linear=True without controls, _compute_per_period_dids
    and _compute_cohort_recentered_inputs now use raw y_pivot outcomes
    (not the first-differenced Z_mat). Previously the per-period path
    would double-difference the already-differenced outcomes.

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

Author: igerber

diff_diff/chaisemartin_dhaultfoeuille.py

@@ -1147,7 +1147,9 @@ def fit(
             # path (L_max >= 1). The per-period path uses binary
             # joiner/leaver categorization and is not part of the DID^X
             # contract (Web Appendix Section 1.2).
-            Y_mat=Y_mat_raw if controls is not None else Y_mat,
+            # Use raw outcomes for per-period DID when controls or
+            # trends_linear is active (both transform Y_mat).
+            Y_mat=Y_mat_raw if controls is not None else (y_pivot.to_numpy() if _is_trends_linear else Y_mat),
             N_mat=N_mat_orig,
             periods=all_periods,
         )
@@ -1581,7 +1583,8 @@ def fit(
         ) = _compute_cohort_recentered_inputs(
             D_mat=D_mat,
             # Phase 1 IF uses per-period structure: use raw outcomes
-            Y_mat=Y_mat_raw if controls is not None else Y_mat,
+            # when controls or trends_linear transform Y_mat.
+            Y_mat=Y_mat_raw if controls is not None else (y_pivot.to_numpy() if _is_trends_linear else Y_mat),
             N_mat=N_mat_orig,
             n_10_t_arr=n_10_t_arr,
             n_00_t_arr=n_00_t_arr,