Fix fast image processor warnings in object detection examples (#35892)

Have the DETR examples default to using the fast image  processor

examples/pytorch/object-detection/run_object_detection.py

@@ -271,6 +271,10 @@ class DataTrainingArguments:
             )
         },
     )
+    use_fast: Optional[bool] = field(
+        default=True,
+        metadata={"help": "Use a fast torchvision-base image processor if it is supported for a given model."},
+    )
 
 
 @dataclass
@@ -427,6 +431,7 @@ def main():
         size={"max_height": data_args.image_square_size, "max_width": data_args.image_square_size},
         do_pad=True,
         pad_size={"height": data_args.image_square_size, "width": data_args.image_square_size},
+        use_fast=data_args.use_fast,
         **common_pretrained_args,
     )