[Quant] Qlinear Add qint8 support backed by xnnpack (#73672)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/73672
If built w/ USE_XNNPACK, for qint8 dtype, qlinear is performed using xnnpack. For quint8 input dtype, use QNNPACK.
Reviewed By: kimishpatel
Differential Revision: D34022360
fbshipit-source-id: c597a670df16e3012e0f78b27184bf8b832e12d9
(cherry picked from commit 971cc36ea1c1575d30061921449e2109f5e4f0a1)