tolist() support for FunctionalTensor

Summary:
Support tolist() for FunctionalTensor for KJT in torch.export

bypass-github-pytorch-ci-checks

Reviewed By: ezyang

Differential Revision: D53731064

fbshipit-source-id: a226bef5de4cdbf6aa0dcc1a6dee28c0874ac484

Author: PaulZhang12

torchrec/distributed/tests/test_pt2.py

@@ -240,3 +240,18 @@ def test_maybe_compute_kjt_to_jt_dict(self) -> None:
             # TODO: turn on AOT Inductor test once the support is ready
             test_aot_inductor=False,
         )
+
+    def test_tensor_tolist(self) -> None:
+        class M(torch.nn.Module):
+            def forward(self, kjt: KeyedJaggedTensor):
+                return kjt.values().tolist()
+
+        kjt: KeyedJaggedTensor = make_kjt([2, 3, 4, 5, 6], [1, 2, 1, 1])
+        self._test_kjt_input_module(
+            M(),
+            kjt.keys(),
+            (kjt._values, kjt._lengths),
+            test_dynamo=False,
+            test_aot_inductor=False,
+            test_pt2_ir_export=True,
+        )