Use cuda-nvtx.

[bdice]

Checklist

• [x] Used a personal fork of the feedstock to propose changes
• [x] Bumped the build number (if the version is unchanged)
• [x] Reset the build number to 0 (if the version changed)
• [x] Re-rendered with the latest conda-smithy (Use the phrase code>@<space/conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• [x] Ensured the license file is being packaged.

This PR uses the new package cuda-nvtx instead of nvtx-c as the source of the NVTX headers. We ignore the run_exports of that package, to avoid introducing a __cuda metapackage dependency on the driver version. This nvtx package is independent of CUDA version and supports CPU-only as well.

[conda-forge-webservices[bot]]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

[bdice]

Closing this. After some offline discussion with @jakirkham we no longer think this rebuild is necessary. It is fine to use nvtx built on nvtx-c alongside cuda-nvtx. The packages won't clobber files because cuda-nvtx puts its contents into a targets/ directory for CUDA, and nvtx-c only provides header-only components so there can't be library conflicts.