[enhancement] expose original fast DAAL interface for self data distances in the oneDAL cosine_distances algo

[icfaust]

Description

Questions have been raised as to the new cosine distances implementation for speed and optimization. In order to speed changes in sklearnex (uxlfoundation/scikit-learn-intelex#2657) we need to open up the original optimized code path for cpu data. The optimization of two input distances can then be improved at a later input. The goal would be to add changes in sklearnex side to limit oneDAL offloading to single input data to match the daal4py implementation. These changes are needed in oneDAL to add the necessary interfaces.

It does it by removing y.has_data() errors and instead piping them to the proper pathways. Note that this means we are still using a somewhat unoptimized GPU path.

---

Checklist:

Completeness and readability

• I have commented my code, particularly in hard-to-understand areas.
• I have updated the documentation to reflect the changes or created a separate PR with updates and provided its number in the description, if necessary.
• Git commit message contains an appropriate signed-off-by string (see CONTRIBUTING.md for details).
• I have resolved any merge conflicts that might occur with the base branch.

Testing

• I have run it locally and tested the changes extensively.
• All CI jobs are green or I have provided justification why they aren't.
• I have extended testing suite if new functionality was introduced in this PR.

Performance

• I have measured performance for affected algorithms using scikit-learn_bench and provided at least a summary table with measured data, if performance change is expected.
• I have provided justification why performance and/or quality metrics have changed or why changes are not expected.
• I have extended the benchmarking suite and provided a corresponding scikit-learn_bench PR if new measurable functionality was introduced in this PR.

[icfaust]

/intelci: run

[icfaust]

/intelci: run

[icfaust]

/intelci: run