Add slicing and int_repr() to QTensor (#19296)
Summary:
Stack:
:black_circle: **#19296 [pt1][quant] Add slicing and int_repr() to QTensor** [:yellow_heart:](https://our.intern.facebook.com/intern/diff/
D14756833/)
:white_circle: #18960 [pt1][quant] Add empty_quantized [:yellow_heart:](https://our.intern.facebook.com/intern/diff/
D14810261/)
:white_circle: #19312 Use the QTensor with QReLU [:yellow_heart:](https://our.intern.facebook.com/intern/diff/
D14819460/)
:white_circle: #19319 [RFC] Quantized SumRelu [:yellow_heart:](https://our.intern.facebook.com/intern/diff/
D14866442/)
Methods added to pytorch python frontend:
- int_repr() returns a CPUByte Tensor which copies the data of QTensor.
- Added as_strided for QTensorImpl which provides support for slicing a QTensor(see test_torch.py)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/19296
Differential Revision:
D14756833
Pulled By: jerryzh168
fbshipit-source-id:
6f4c92393330e725c4351d6ff5f5fe9ac7c768bf