feat(gpu): implement CUDA Graph to accelerate default classical and multibit PBS

[pdroalves]

closes: please link all relevant issues

PR content/description

Check-list:

• Tests for the changes have been added (for bug fixes / features)
• Docs have been added / updated (for bug fixes / features)
• Relevant issues are marked as resolved/closed, related issues are linked in the description
• Check for breaking changes (including serialization changes) and add them to commit message following the conventional commit specification

[agnesLeroy]

Hey! Seems like the PBS test is failing 🤔 we can talk about it during the sync 🙂

[pdroalves]

@agnesLeroy I was using CUDA Graphs with cudaStreamCaptureModeGlobal, which basically forbids CPU multi-threading. I replaced it by a more strict mode that records only things happening on a single stream. That seems to fix the classical PBS issue. There is still something going on with the multi-bit PBS. I will check it tomorrow.