[PyTorch Edge] Clean up Quantize Tensor code (#66220)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/66220
- Pass pointers rather than tensors to ```quantize_tensor_arm``` to allow for using ```__restrict__``` and to make parallelization easier (as in the next diff on this stack D31205883)
- Replace ```auto``` with actual types
- Replace raw cast with reinterpret_cast<...>
- All of these changes make the code structure similar to that of Dequantize
ghstack-source-id: 142166376
Test Plan: same as D31066997 (all tests pass)
Reviewed By: kimishpatel
Differential Revision: D31444248
fbshipit-source-id: 6a31d090082047263403f415911c199519987595
