[CUDA] Change slim-wheel libraries load order (#145638)

There is no libnvjitlink in  CUDA-11.x , so attempts to load it first will abort the execution and prevent the script from preloading nvrtc

Fixes issues reported in https://github.com/pytorch/pytorch/pull/145614#issuecomment-2613107072

Pull Request resolved: https://github.com/pytorch/pytorch/pull/145638
Approved by: https://github.com/atalman, https://github.com/kit1980, https://github.com/malfet

Co-authored-by: Nikita Shulga <2453524+malfet@users.noreply.github.com>
(cherry picked from commit 2a70de7e9257e3f8c2874a10e3612c8939b79867)

Co-authored-by: Wei Wang <weiwan@nvidia.com>