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

config/settings.py

@@ -75,8 +75,7 @@ class InferenceConfig:
     model_path: str = "./models/yolo11n.engine"
     input_width: int = 480
     input_height: int = 480
-    batch_size: int = 1
-    max_batch_size: int = 8
+    batch_size: int = 4
     conf_threshold: float = 0.5
     nms_threshold: float = 0.45
     device_id: int = 0
@@ -160,8 +159,7 @@ class Settings:
             model_path=os.getenv("MODEL_PATH", "./models/yolo11n.engine"),
             input_width=int(os.getenv("INPUT_WIDTH", "480")),
             input_height=int(os.getenv("INPUT_HEIGHT", "480")),
-            batch_size=int(os.getenv("BATCH_SIZE", "1")),
-            max_batch_size=int(os.getenv("MAX_BATCH_SIZE", "8")),
+            batch_size=int(os.getenv("BATCH_SIZE", "4")),
             conf_threshold=float(os.getenv("CONF_THRESHOLD", "0.5")),
             nms_threshold=float(os.getenv("NMS_THRESHOLD", "0.45")),
         )