Make empty_affine_quantized private (#19446)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/19446
change empty_affine_quantized to _empty_affine_quantized
Reviewed By: dzhulgakov
Differential Revision: D15008757
fbshipit-source-id: c7699ac0c208a8f17d88e95193970c75ba7219d3
