nn_quantized test (#23169)
Summary:
- scale/zero_point in quantized modules should be Tensor
- fix conv module permutation API
Pull Request resolved: https://github.com/pytorch/pytorch/pull/23169
ghstack-source-id: 86956383
Reviewed By: zafartahirov
Differential Revision: D16423570
fbshipit-source-id: d29498e07bdd8f71a33b4e16e089f80847bbca6d