Create CUDA-aware futures in RequestCallback (#58426)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/58426
The operations in RequestCallback can return CUDA tensors, thus the futures used to hold them must be CUDA-aware.
ghstack-source-id: 129567051
Test Plan: CI
Reviewed By: mrshenli
Differential Revision: D28474981
fbshipit-source-id: 492b8e71a43da5f63b4b7a31f820427cde9736e4