feat: hgb-v2 — regression heads predict rise, recovering flood warning lead

Rolling-origin evaluation (5 monsoon folds x 4 variants, P.1 + P.103;
results in models/eval_variants.json) showed the absolute-level target
alerting AT the crossing on essentially every event, while the rise
target (future max - current level, level added back at serving) gives
+6h on the hard 2024 crossings, +45h in 2025, fewer false alarms than
weighted/quantile variants, and ~11% better MAE. Weighted and quantile
variants rejected: more false alarms, no Brier-score calibration gain.

Ported to production: train.py fits rise in both eval and refit passes
(sigma/metrics computed in absolute space), bundles stamped hgb-v2 with
regression_target='rise', predict.py adds the level back for v2 and
stays compatible with v1 bundles, backtest_render.py mirrors the same
math. Regenerated backtest charts: 2024 first alert 11:00 24 Sep (6h
BEFORE the 17:00 crossing, was 18h after), 2025 alert 45h ahead, and
the record-peak underprediction is gone (rise models can exceed the
training max). The >=12h acceptance gate still fails honestly at +6h —
closing that needs rainfall inputs. New P.1 MAE 5.0/7.2/9.4 cm at
6/12/24h; docs updated throughout.

src/ml/predict.py

@@ -161,7 +161,11 @@ def _model_forecast(
         if reg is None:
             results.append(None)
             continue
-        predicted_max = max(float(reg.predict(feature_row)[0]), current_level)
+        raw_prediction = float(reg.predict(feature_row)[0])
+        if bundle.get("regression_target") == "rise":
+            # v2 bundles predict the rise over the current level
+            raw_prediction += current_level
+        predicted_max = max(raw_prediction, current_level)
         sigma_h = bundle["sigma"].get(horizon_h, HEURISTIC_SIGMA)
 
         # Belt-and-braces: the classifier head OR the regression-sigmoid path,