solved serde issue when serializing op using list of faketensor as output

Differential Revision: D82473504

Pull Request resolved: pytorch#14308

Author: Gasoonjia

exir/serde/export_serialize.py

@@ -1211,7 +1211,7 @@ def serialize_hoo_outputs(self, node: torch.fx.Node) -> List[Argument]:
         """
         meta_val = node.meta["val"]
 
-        if isinstance(meta_val, tuple):
+        if isinstance(meta_val, (tuple, list)):
             # Note: Since we don't have a schema, we just serialize all tuple
             # outputs to be a list of values. Even if the output is supposed to
             # be a tensor list (Tensor[]), we will serialize it to be a list of