fix: ML Pipeline 缺失值填充 + 推理端外部模型 dict 兼容

- modeling_batch: Pipeline 首步新增 SimpleImputer(median) 填充 NaN
- inference_batch: 外部模型支持完整 dict（含 metadata/train_wavelengths），
  兼容旧版裸 Pipeline 对象
- step9_ml_predict_panel: 模型加载保留完整 dict 而非仅 model 对象，
  确保推理端可从 train_wavelengths 做光谱重采样

src/core/modeling/modeling_batch.py

@@ -20,6 +20,7 @@ from sklearn.ensemble import GradientBoostingRegressor, AdaBoostRegressor, Extra
 from sklearn.tree import DecisionTreeRegressor
 from sklearn.neural_network import MLPRegressor
 from sklearn.pipeline import Pipeline
+from sklearn.impute import SimpleImputer
 from joblib import parallel_backend
 # 第三方模型导入
 # try:
@@ -611,6 +612,7 @@ class WaterQualityModelingBatch:
         # ============ 关键：把预处理器塞进 Pipeline ============
         preproc = get_preprocessing_transformer(preprocess_method)
         pipeline = Pipeline([
+            ('imputer', SimpleImputer(strategy='median')),
             ('preproc', preproc),
             ('model', base_model),
         ])