✨[Feature] Automate the release process

[lanluo-nvidia]

Is your feature request related to a problem? Please describe. Currently when we do a release, there are following manual steps to run on the local workstation:

1. 8 pip packages which require to upload to pypi via twine: 1.1) build 4 pip packages by auditwheel for linux on cuda12.1 for python 3.8/3.9/3.10/3.11 1.2) build 4 pip packages for windows on cuda12.1 for python 3.8/3.9/3.10/3.11

2)4 tarballs for c++ libraries for pre-cxx11-abi and cxx11-abi for cuda11.8 + cuda12.1

3)valida on all the pip wheel file built locally

These are time consuming and also repeated every time when we cut a release candidate

Describe the solution you'd like We need to be able to automate the release procedures as much as we can

Plan on adding a release workflow which will do all the manual steps listed above in the CI/CD

Describe alternatives you've considered

Additional context