Add feature flag for mm inputs processing optimization (#13278)

Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>

python/sglang/srt/server_args.py

@@ -565,6 +565,7 @@ class ServerArgs:
     # For Multi-Modal
     mm_max_concurrent_calls: int = 32
     mm_per_request_timeout: float = 10.0
+    enable_broadcast_mm_inputs_process: bool = False
 
     # For checkpoint decryption
     decrypted_config_file: Optional[str] = None
@@ -3656,6 +3657,12 @@ class ServerArgs:
             default=ServerArgs.mm_per_request_timeout,
             help="The timeout for each multi-modal request in seconds.",
         )
+        parser.add_argument(
+            "--enable-broadcast-mm-inputs-process",
+            action="store_true",
+            default=ServerArgs.enable_broadcast_mm_inputs_process,
+            help="Enable broadcast mm-inputs process in scheduler.",
+        )
 
         # For checkpoint decryption
         parser.add_argument(