run tests on CUDA 13

[jameslamb]

Contributes to rapidsai/build-planning#208

Proposes running tests in CUDA 13 environments. jupyterlab-nvdashboard is pure-Python and shouldn't be dependent on CUDA version, but doing this will give us higher confidence that its dependencies like pynvml are working correctly across both major CUDA versions supported by RAPIDS.

Notes for Reviewers

This switches GitHub Actions workflows to the cuda13.0 branch from here: rapidsai/shared-workflows#413

A future round of PRs will revert that back to branch-25.10, once all of RAPIDS supports CUDA 13.

[jameslamb]

Sorry, realizing that these filter changes are resulting in only CUDA 13 test coverage. I want to fiddle with this a bit to get one CUDA 12 test job and one CUDA 13 test job.

[jameslamb]

Blegh, how many times do I have to trip over "you can only test on the combinations of CUDA, Python, and CPU arch you build for, even for pure-Python packages" before I remember 😅

Downloading and decompressing jupyterlab-nvdashboard_conda_python_cuda12_py312_aarch64 from Run ID 17144910033 into /tmp/tmp.NuGn5WmiEv
no artifact matches any of the names or patterns provided

(build link)

ref: rapidsai/build-planning#43

I'm going to revert this back to the state that was already approved (just testing on CUDA 13) and merge it. This project isn't receiving a lot of updates and the CUDA 12 support has been thoroughly tested, so I think it's safe to switch to testing only on CUDA 13 for now, to give us higher confidence in the CUDA 13 rollout for RAPIDS.

[jameslamb]

/merge