[diffusion] chore: improve the output_path config and enable the server to return inference duration (#16965)

Co-authored-by: Xiaoyu Zhang <35585791+BBuf@users.noreply.github.com>

python/sglang/multimodal_gen/runtime/server_args.py

@@ -328,6 +328,8 @@ class ServerArgs:
 
     scheduler_port: int = 5555
 
+    output_path: str | None = None
+
     # Prompt text file for batch processing
     prompt_file_path: str | None = None
 
@@ -682,6 +684,12 @@ class ServerArgs:
             default=ServerArgs.webui_port,
             help="Whether to use webui for better display",
         )
+        parser.add_argument(
+            "--output-path",
+            type=str,
+            default=ServerArgs.output_path,
+            help="Directory path to save generated images/videos",
+        )
 
         # LoRA
         parser.add_argument(