feat: TensorRT 固定 batch=4 重构

- tensorrt_engine.py 工业级 Buffer Pool
- preprocessor.py 添加 pad_to_batch4()
- postprocessor.py 支持批量输出
- settings.py 固定 batch_size=4

main.py

@@ -204,7 +204,7 @@ class EdgeInferenceService:
         frame: VideoFrame,
         roi
     ):
-        """处理ROI帧"""
+        """处理ROI帧，固定 batch=4 推理"""
         try:
             if not roi.enabled:
                 return
@@ -213,7 +213,7 @@ class EdgeInferenceService:
             
             processed_image, scale_info = cropped
             
-            batch_data = self._preprocessor._batch_preprocessor._stack_and_normalize(
+            batch_data, _ = self._preprocessor._batch_preprocessor.preprocess_batch(
                 [processed_image]
             )