[torch.compile] use empty tensor instead of None for profiling

[torch.compile] use empty tensor instead of None for profiling #16691