[PyTorch Edge] Perform QMatMul Requantization within Ruy (#73245)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/73245
The multiplier fixedpoint and multiplier exponent as determined by ruy::QuantizeMultiplier are explained in section 2.2 of https://arxiv.org/pdf/1712.05877.pdf
ghstack-source-id: 151250872
Test Plan:
From fbcode:
```buck test caffe2/test:quantization -- test_qmatmul```
Performance Improvement/Benchmarking: See testplan of D34012771
Reviewed By: kimishpatel
Differential Revision: D33733271
fbshipit-source-id: 031131fd49e139629734e82f21977b147016dc73
(cherry picked from commit 592123974db93eea634391fdfe75dad9b9c00ca1)
