fix: DualStream_MNF 推理重建113列后交给 Pipeline.predict() 完整处理

- 之前手动调 imputer/preproc.transform 后返回， predict() 又套一层
  Pipeline → SimpleImputer 收到44列不匹配113列
- 改为重建113列训练特征集，返回后让 model.predict() 完整走
  Pipeline: imputer→preproc→cleaner→SVR，与训练完全一致

src/core/prediction/inference_batch.py

@@ -477,17 +477,15 @@ class WaterQualityInference:
         metadata = self.loaded_model_data.get('metadata', {})
 
         # ═══════════════════════════════════════════════════════════
-        # ★ DualStream_MNF 快速通道：Pipeline 已内置 MNF + 物理指数
-        #   但需先将推理光谱重采样到训练特征空间（308→113列），
-        #   再走 Pipeline 的 imputer+preproc 步骤
+        # ★ DualStream_MNF：重建训练特征空间（308→113列），
+        #   让 Pipeline.predict() 完整走 imputer→preproc→cleaner→SVR
         # ═══════════════════════════════════════════════════════════
         if actual_preprocess_method == "DualStream_MNF" and isinstance(model, Pipeline):
-            print("[DualStream_MNF] 推理：先重建训练特征集再直通 Pipeline")
-            # 1) 重采样 308 波段 → 训练波长网格
+            print("[DualStream_MNF] 推理：重建训练特征集，交给 Pipeline 完整处理")
             train_wl = metadata.get('train_wavelengths', None)
             train_cols = metadata.get('train_columns', None)
             if train_wl is not None and len(train_wl) > 0 and train_cols is not None:
-                # 从 spectra 中提取光谱列（列名可转 float 的即为波长列）
+                # 1) 重采样 308 → 训练波长
                 spec_cols = []
                 for c in spectra.columns:
                     try:
@@ -502,34 +500,27 @@ class WaterQualityInference:
                     resampled[i] = np.interp(dst_wl, src_wl, spec_data[i],
                                              left=np.nan, right=np.nan)
                 resampled = np.nan_to_num(resampled, nan=0.0)
-                # 2) 计算 WQI 补齐到训练列数
                 wl_col_names = [f'{wl:.6f}' for wl in train_wl]
-                spectra_df = pd.DataFrame(resampled, columns=wl_col_names)
+                spectra = pd.DataFrame(resampled, columns=wl_col_names)
+                # 2) 计算 WQI + 补零到训练列数
                 n_need = len(train_cols)
-                if spectra_df.shape[1] < n_need:
-                    print(f"[DualStream_MNF] 计算 WQI 补齐: {spectra_df.shape[1]} → {n_need}")
+                if spectra.shape[1] < n_need:
+                    print(f"[DualStream_MNF] 计算 WQI: {spectra.shape[1]} → {n_need}")
                     try:
                         from src.utils.water_index import WaterQualityIndexCalculator
-                        calc = WaterQualityIndexCalculator()
-                        formulas = calc.list_available()
-                        wqi = calc.calculate_many(formulas, spectra_df, fast=True)
-                        spectra_df = pd.concat([spectra_df, wqi], axis=1)
+                        wqi = WaterQualityIndexCalculator().calculate_many(
+                            WaterQualityIndexCalculator().list_available(),
+                            spectra, fast=True)
+                        spectra = pd.concat([spectra, wqi], axis=1)
                     except Exception as e:
                         print(f"[DualStream_MNF] WQI 失败: {e}")
-                    # 如果还不够，补零
-                    while spectra_df.shape[1] < n_need:
-                        spectra_df[f'_pad_{spectra_df.shape[1]}'] = 0.0
-                    spectra_df = spectra_df.iloc[:, :n_need]
-                # 3) 喂入 Pipeline
-                spectra_processed = model.named_steps['imputer'].transform(spectra_df.values)
-                spectra_processed = model.named_steps['preproc'].transform(spectra_processed)
-            else:
-                spectra_processed = model.named_steps['imputer'].transform(spectra.values)
-                spectra_processed = model.named_steps['preproc'].transform(spectra_processed)
-            spectra_processed = np.nan_to_num(spectra_processed, nan=0.0, posinf=0.0, neginf=0.0)
-            print(f"[DualStream_MNF] 预处理完成: {spectra_processed.shape}")
-            print(f"[特征对齐] 最终输入维度: {spectra_processed.shape}")
-            return spectra_processed
+                    while spectra.shape[1] < n_need:
+                        spectra[f'_pad_{spectra.shape[1]}'] = 0.0
+                    spectra = spectra.iloc[:, :n_need]
+            print(f"[DualStream_MNF] 特征重建完成: {spectra.shape}")
+            print(f"[特征对齐] 最终输入维度: {spectra.shape}")
+            # 返回 DataFrame，后续 model.predict() 走完整 Pipeline
+            return spectra.values
 
         train_wavelengths = metadata.get('train_wavelengths', None)
         # 旧模型无 train_wavelengths → 不做波长匹配，走下方分支 B 的 linspace 路径