pytorch 2: TypedStorage is deprecated

[rijobro]

Describe the bug

With pytorch 2, I'm getting lots of instances of this warning:

/home/rbrown/miniconda3/envs/py3.11/lib/python3.11/site-packages/torch/_tensor.py:1287: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()
ret = func(*args, **kwargs)

The only usage of torch.storage() is here, so I suspect that's the problem. @wyli I see you added the code, can we do without it or modify it?

[rijobro]

Can I use tensor.untyped_storage instead?

[wyli]

I basically followed this method

https://github.com/pytorch/pytorch/blob/v1.13.1/torch/multiprocessing/reductions.py#L149-L150

and the expected behaviour is mainly tested here

MONAI/tests/test_meta_tensor.py

Lines 523 to 536 in 6803061

	def test_multiprocessing(self, device=None, dtype=None):
	"""multiprocessing sharing with 'device' and 'dtype'"""
	buf = io.BytesIO()
	t = MetaTensor([0, 0] if dtype in (torch.int32, torch.int64) else [0.0, 0.0], device=device, dtype=dtype)
	t.is_batch = True
	if t.is_cuda:
	with self.assertRaises(NotImplementedError):
	ForkingPickler(buf).dump(t)
	return
	ForkingPickler(buf).dump(t)
	obj = ForkingPickler.loads(buf.getvalue())
	self.assertIsInstance(obj, MetaTensor)
	assert_allclose(obj.as_tensor(), t)
	assert_allclose(obj.is_batch, True)

ForkingPickler can dump and load metatensors.

feel free to update it for the latest API