[iOS GPU][Perf][4/n] Reuse the same command buffer when copying results to CPU (#57667)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/57667
Context - https://fb.workplace.com/groups/pytorch.edge.team/permalink/855194118368662/
Got 5% win for mobilenetv2 and unet
ghstack-source-id: 128338532
Test Plan: - CI
Reviewed By: kimishpatel
Differential Revision: D28116806
fbshipit-source-id: b9c766c58ae41f3408724ec962695f38985ace05