Implement linear module for APoT quantization (#82105)
### Summary
Implement linear module to support APoT quantization. Use bitshifting method discussed in APoT paper https://arxiv.org/pdf/1909.13144.pdf to multiply PoT terms in APoT quantized weight tensor with uniformly quantized activation tensor to demonstrate alternative to matrix multiplication.
Multiplication using bitshifting for PoT:
<img width="340" alt="Screen Shot 2022-07-25 at 12 44 26 PM" src="https://user-images.githubusercontent.com/68875504/180831050-ff849bca-8eb0-4b69-9b7f-c6c94a4cdfb5.png">
### Test Plan
Run unit tests with: `python /pytorch/test/quantization/core/experimental/test_linear.py`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/82105
Approved by: https://github.com/HDCharles