[fx2trt] Use itensor_to_tensor_meta to track the TensorMeta info for ITensor node (#65427)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/65427
Previously we added a input_tensor_meta for dequantize function, this is a bit hacky since this creates a dependency between
the arguments of dequantize and if there are passes that changes the input then we would need to update tensor meta as well
Test Plan:
python torch/fx/experimental/fx2trt/example/quantized_resnet_test.py
Imported from OSS
Reviewed By: soulitzer
Differential Revision: D31094274
fbshipit-source-id: 5e40648d3081e2363f3a70bcc9745df4a8190ad3