Inductor: fix incorrect result of inplace unsqueeze (#94797)
This pr aims to fix the incorrect result in the following test case.
```
@torch._dynamo.optimize("inductor")
def fn(a):
unsqueeze_ = torch.ops.aten.unsqueeze_.default(a, 0)
return unsqueeze_
args = [
((1, 1, 1, 12, 11, 3), (396, 396, 396, 33, 3, 1), torch.int64, "cpu")
]
args = [rand_strided(sh, st, dt, dev) for (sh, st, dt, dev) in args]
with torch.no_grad():
out = fn(*args)
# expected result: (396, 396, 396, 396, 33, 3, 1) torch.Size([1, 1, 1, 1, 12, 11, 3])
print(args[0].stride(), args[0].shape) # incorrect result: (396, 396, 396, 396, 396, 396, 33, 3, 1) torch.Size([1, 1, 1, 1, 1, 1, 12, 11, 3])
```
**Root cause**
1. [fake_tensor](https://github.com/pytorch/pytorch/blob/master/torch/_dynamo/variables/builder.py#L140) is changed during [tracer.run](https://github.com/pytorch/pytorch/blob/master/torch/_dynamo/convert_frame.py#L311), then it will [pass incorrect inputs to inductor](https://github.com/pytorch/pytorch/blob/master/torch/_dynamo/output_graph.py#L670).
2. example_inputs are changed during [propagate](https://github.com/pytorch/pytorch/blob/master/torch/_inductor/mkldnn.py#L509)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/94797
Approved by: https://github.com/jgong5, https://github.com/jansel
