docs(timeseries): document lpirfs_cholesky in agent card + reference

Surfaces the new identification='lpirfs_cholesky' path in the
local_projections agent card (assumption, pre-condition, and the
horizon-0-zero Cholesky-ordering failure mode) and fixes the reference
example to the real signature (outcome=, horizons int, auto_lag) plus a
unit-Cholesky-shock snippet. Pairs with the local_projections feature.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: brycewang-stanford

docs/reference/timeseries.md

@@ -48,10 +48,18 @@ sp.cointegration(..., method='phillips_ouliaris')
 
 ```python
 sp.local_projections(
-    df, y='gdp', shock='mp_shock',
-    horizons=range(0, 21),
-    controls=['infl_lag','r_lag'],
-    se='driscoll_kraay',                            # HAC-panel SE
+    df, outcome='gdp', shock='mp_shock',
+    horizons=20,
+    controls=['infl_lag', 'r_lag'],
+    auto_lag=False,                                 # controls are used verbatim
+)
+
+# Match lpirfs::lp_lin with a unit Cholesky shock.
+sp.local_projections(
+    df, outcome='gdp', shock='mp_shock',
+    horizons=20,
+    identification='lpirfs_cholesky',
+    endog_order=['gdp', 'mp_shock'],
 )
 ```
 

src/statspai/_agent_cards_extra.py

@@ -1867,10 +1867,16 @@
             "Controls are passed at their time-t values verbatim "
             "— lag them yourself; the estimator does not re-lag "
             "them",
+            "For identification='lpirfs_cholesky', the endogenous "
+            "variable order defines the recursive structural shock, "
+            "matching lpirfs::lp_lin with shock_type=1",
         ],
         "pre_conditions": [
             "Single time-ordered series in a DataFrame with "
             "the outcome and shock columns",
+            "For identification='lpirfs_cholesky', endog_order "
+            "contains the outcome and shock columns in the intended "
+            "Cholesky ordering",
             "Length comfortably exceeds horizons + max lag so "
             "the deepest horizon regression retains enough "
             "usable rows",
@@ -1888,6 +1894,15 @@
                 "columns are not formed",
                 "alternative": "var",
             },
+            {
+                "symptom": "The horizon-0 response is zero when the "
+                "shock is ordered after the response variable",
+                "exception": "(none — Cholesky ordering convention)",
+                "remedy": "Use the intended endog_order, or switch "
+                "back to identification='direct' if the shock column "
+                "is already externally identified",
+                "alternative": "var",
+            },
             {
                 "symptom": "Series too short: the "
                 "longest-horizon regression has "