Why the upper_bound of qrange_len is set to 2**31 when dtype is torch.qint32?
https://github.com/pytorch/pytorch/blob/2628901033e7df5b87a9b7137ddffb450d4a7c42/torch/ao/quantization/utils.py#L356
This looks like mismatch with the following code:
https://github.com/pytorch/pytorch/blob/2628901033e7df5b87a9b7137ddffb450d4a7c42/torch/ao/quantization/utils.py#L374