Update DLPack to v1.0

[alexbarghi-nv]

Summary

• require DLPack >=1.0,<1.1 in dependencies.yaml
• update the CMake fallback tag/minimum version to v1.0
• regenerate all checked-in conda environment files with rapids-dependency-file-generator 1.21.0

Impact

cuGraph builds and downstream development environments now consistently use the DLPack 1.0 release series, while avoiding unvalidated changes from 1.1 and later.

Release target

This PR now targets main for the 26.10 release. It was retargeted from release/26.08 so the DLPack 1.x migration can be coordinated across RAPIDS projects outside the 26.08 code freeze. The cross-project support plan is tracked in rapidsai/build-planning#308.

Validation

• reran rapids-dependency-file-generator and verified the output is idempotent
• parsed dependencies.yaml and all generated all_*.yaml environment files with PyYAML
• ran git diff --check
• verified all generated conda environments use the >=1.0,<1.1 constraint

Constraints

• coordinate with Support DLPack 1.0 build-planning#308 before merging

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[alexbarghi-nv]

/ok to test e66cbff

[copy-pr-bot]

/ok to test e66cbff

@alexbarghi-nv, there was an error processing your request: E2

See the following link for more information: https://docs.gha-runners.nvidia.com/cpr/e/2/

[alexbarghi-nv]

/ok to test e55cbff

[alexbarghi-nv]

/ok to test 7550843

[alexbarghi-nv]

/ok to test 7760d21

[alexbarghi-nv]

/ok to test 7d0fee5

[bdice]

This is blocked without a DLPack update in cuDF.

error    libmamba Could not solve for environment specs
    The following packages are incompatible
    ├─ cuda-version =13.3 * is requested and can be installed;
    ├─ dlpack >=1.0,<1.1 * is requested and can be installed;
    └─ libcudf =26.8,>=0.0.0a0 * is not installable because there are no viable options
       ├─ libcudf [26.08.00a943|26.08.00a944|...|26.08.00a988] would require
       │  └─ cuda-version >=12,<13.0a0 *, which conflicts with any installable versions previously reported;
       └─ libcudf [26.08.00a943|26.08.00a944|...|26.08.00a988] would require
          └─ dlpack >=0.8,<1.0 *, which conflicts with any installable versions previously reported.

However, cuDF is frozen for 26.08. The version discrepancy is blocking our CCCL-RAPIDS CI, which is one of the places where we need to support "unified" environments with all of RAPIDS.

We need to revert NVIDIA/cuvs#2354 for 26.08 and retarget this DLPack update to 26.10. As a part of that, we also need to make sure we are following DLPack practices for version stability:

Starting DLPack 1.0, the DLManagedTensorVersioned struct should be used and the existing DLManagedTensor struct is considered deprecated, though a library should try to support both during the transition period if possible.

For the future, it would be helpful to document the motivation for cross-library dependency changes like this with an issue in https://github.com/rapidsai/build-planning/issues. The motivation for this update is not specified in this PR or in the cuVS PR.

[alexbarghi-nv]

Superseded by #5590, which now includes the DLPack v1.0 CMake fallback, the dlpack>=1.0,<1.1 dependency constraint, and regenerated 26.10 environment files.