Use CCCL v2.7.0 final tag.

[bdice]

Description

CCCL is currently pinned to v2.7.0-rc2. Now that a final tag for v2.7.0 is available, we should use that instead.

See diff: NVIDIA/cccl@v2.7.0-rc2...v2.7.0

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.
• The cmake-format.json is up to date with these changes.
• I have added new files under rapids-cmake/
  • I have added include guards (include_guard(GLOBAL))
  • I have added the associated docs/ rst file and update the api.rst

[bdice]

From reviewing the diff, I do not anticipate any risk from merging this as-is, without needing to do test PRs for all of RAPIDS. @robertmaynard What do you think?

[bdice]

I’ll merge this around the same time as the GCC 13 upgrades (#741) so that we only bust the RAPIDS sccache one time.

[PointKernel]

CCCL just merged a PR that enables progress on relaxed constexpr work in cuDF: NVIDIA/cccl#3382

Could we include that change in this PR as well? I was planning to open a PR to update the CCCL version in rapids-cmake, but I noticed this PR and wanted to check here first.

[bdice]

@PointKernel We should move to a stable tag first. Can we defer on a future upgrade until 25.04? CCCL RAPIDS builds are currently failing and may be broken for a bit. We can’t adopt new CCCL versions with confidence until the CCCL RAPIDS CI is fixed.

[PointKernel]

Can we defer on a future upgrade until 25.04?

Sure, no problem.

[bdice]

/merge