Support CUDA 13

[mitchdz]

CUDA 13 Support

This update adds CUDA 13.0 build support.

Working deployments test - https://github.com/NVIDIA/cuda-quantum/actions/runs/18562671783

Changes

• Add support for CUDA 13, including CC 100 (Blackwell) only when CUDA ≥ 13.0
  • CUDA 12.6 does not recognize CC 100; future updates will move to CUDA 12.9
• Remove Volta (CC 70) from CMake targets
• Add Turing (CC75) to CMake targets
• Update docs to remove CUDA 11.x references
• UCX: v1.16.0 → v1.19.0
• CuPy: 13.4.1 → 13.6.0 for both cupy-cuda12x and cupy-cuda13x
• Default build targets CUDA 13 in pyproject.toml
• Update wheel Dockerfile to handle CUDA 13 ↔ CUDA 12 deps
• (Thanks @1tnguyen) Removed deprecated Thrust calls per CCCL v3.0.0
• prevent libc++ from redefining conflicting headers from CCCL 1934d3a
  • functions are redefined in the new CCL in https://github.com/NVIDIA/cccl/blob/main/libcudacxx/include/cuda/std/__new/allocate.h which conflict with the definitions from the thrust library.
• fix publishing pipeline
  • for python 3.12 skip the test installation error, as the old cuda-quantum pypi packages for 0.8.0 does not exist for those wheels.
• update mgpu sha for cusvim cuda13 support

Open issues (to be address in future PRs)

• publishing pipeline does not work with notebooks
• install nightly torch for cuda13 support
• better handle CCCL update Revisit how we update from CCCL v2.x to v3.x #3519
• Create 2 pyproject.toml files, pyproject.toml.cu{12,13} Use different pyproject.toml file for each cuda version #3520

[bmhowe23]

This passed here (https://github.com/NVIDIA/cuda-quantum/actions/runs/18546302088), so I will force this merge now.

[github-actions]

CUDA Quantum Docs Bot: A preview of the documentation can be found here.