Update on "[HigherOrderOp] stop erroring out on non-Tensor returns"

If map or autograd.Function have an input that returns a non-Tensor,
then the code just errors out. Instead of erroring out we should graph
break by raising Unsupported so users aren't confused. The better thing
to do is actually support non-Tensor returns but that requires more
work.

Test Plan:
- new tests

cc voznesenskym penguinwu anijain2305 @EikanWang jgong5 @Guobing-Chen @XiaobingSuper zhuhaozhe blzheng @Xia-Weiwen wenzhe-nrv jiayisunx ipiszy chenyang78 aakhundov

[ghstack-poisoned]

test/dynamo/test_autograd_function.py

@@ -344,7 +344,9 @@ def f(x):
 
         self.assertEqual(result, Foo.apply(x))
         self.assertEqual(cnt.frame_count, 1)
-        self.assertEqual(list(torch._dynamo.utils.counters['graph_break'].values()), [1])
+        self.assertEqual(
+            list(torch._dynamo.utils.counters["graph_break"].values()), [1]
+        )
 
     @unittest.expectedFailure
     def test_function_with_bound_free_variable(self):

torch/_dynamo/variables/higher_order_ops.py

@@ -484,10 +484,11 @@ def fixup_branch_inps(graph, add_after, new_args, suffix) -> None:
 
 def non_single_tensor_return_unsupported(api, ret):
     from . import TensorVariable
+
     if not isinstance(ret, TensorVariable):
         raise Unsupported(
-            f"{api} over function that returns something "
-            f"other than one Tensor")
+            f"{api} over function that returns something " f"other than one Tensor"
+        )
 
 
 class MapHigherOrderVariable(TorchHigherOrderOperatorVariable):