[fix] correct logic k-NN algos ```kneighbors()``` call when ```algorithm='brute'``` and fit with GPU

[icfaust]

Description

Calling k-NN algorithms kneighbors() after fitting with GPU and algorithm = 'brute' will assume that it was fit using daal4py, and will fail in yielding kneighbors. This adds corrections in that check, and a test for kneighbors without input.

kneighbors is a bit special because it doesn't require an input to yield numerical results (re-using the fitted X values). The fit can occur in the daal4py backend or onedal backend. The check for predict and kneighbors now will check for which train object was generated, and will use it.

---

Checklist to comply with before moving PR from draft:

PR completeness and readability

• [x] I have reviewed my changes thoroughly before submitting this pull request.
• [x] I have commented my code, particularly in hard-to-understand areas.
• [x] I have updated the documentation to reflect the changes or created a separate PR with update and provided its number in the description, if necessary.
• [x] Git commit message contains an appropriate signed-off-by string (see CONTRIBUTING.md for details).
• [x] I have added a respective label(s) to PR if I have a permission for that.
• [x] I have resolved any merge conflicts that might occur with the base branch.

Testing

• [x] The unit tests pass successfully.
• [x] I have run it locally and tested the changes extensively.

Performance

• [x] I have measured performance for affected algorithms using scikit-learn_bench and provided at least summary table with measured data, if performance change is expected.
• [x] I have provided justification why performance has changed or why changes are not expected.