Add nvjitlink to cuda-toolkit pip extras for cusparse compatibility

[bdice]

Summary

• Add nvjitlink to cuda-toolkit[...] extras in dependencies.yaml to ensure pip installs a version-compatible nvidia-nvjitlink alongside nvidia-cusparse
• Without this, pip may resolve an older nvidia-nvjitlink-cu12 than what nvidia-cusparse-cu12 requires, causing undefined symbol: _nvJitLinkGetErrorLogSize_12_9 at runtime

xref: rapidsai/build-planning#253

Summary by CodeRabbit

• Chores
  • Updated CUDA toolkit dependencies to include nvidia-nvjitlink for CUDA 12.* and 13.* versions and default configurations.

[coderabbitai]

📝 Walkthrough

Walkthrough

The pull request adds nvidia-nvjitlink as a new dependency to CUDA toolkit package configurations across multiple files and CUDA versions (12.* and 13.*), alongside existing packages like nvtx, cublas, and cudart.

Changes

Cohort / File(s)	Summary
CUDA toolkit dependencies dependencies.yaml	Added nvidia-nvjitlink to cuda-toolkit package selectors for CUDA 12.* and 13.* in cuda_wheels matrices, and to the general unsuffixed default packages list.
Python libcuopt project configuration python/libcuopt/pyproject.toml	Added nvidia-nvjitlink to the project dependencies list, positioned between nvidia-cusparse and nvidia-nvtx.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main change: adding nvjitlink to cuda-toolkit pip extras for cusparse compatibility, which matches the core modification across dependencies.yaml and pyproject.toml files.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

No actionable comments were generated in the recent review. 🎉

Tip

Issue Planner is now in beta. Read the docs and try it out! Share your feedback on Discord.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[bdice]

/merge