ENH: adding array-api-compat and enabling array api conformance tests

[samir-nasibli]

Description

Adding array-api-compat and enabling array api conformance tests.

Done:

• added array-api-compat to test env.
  • test statistics:
    • 172 Array API tests:
      • SKIPPED 142
        • 129(due to torch/array_api.cupy availability) - not required.
        • 6 due to scikit-learn-intelex functional support doesn't support attributes consistency for the inputs. Disabled now.
          • Expected that real Array API support for PCA or Ridge will resolve this. No ticket required
        • 3 train_test_split inconsistency for Array API inputs.
          • Ticket N8533
        • 1 Ridge regression. Array API functional support supported for all solvers. Not raising error for non-svd solvers.
          • Expected that real Array API support for Ridge will resolve this. No ticket required
        • 1 PCA. Array API functionally supported for all factorizations. power_iteration_normalizer=["LU", "QR"]
        • 2 PCA. InvalidParameterError: The 'M' parameter of randomized_svd must be an instance of 'numpy.ndarray' or a sparse matrix.
          • Ticket N8534
      • PASSED 27

# scikit-learn-intelex functional support doesn't support attributes consistency for the inputs. Disabled now.
SKIPPED decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='covariance_eigh')-check_array_api_input_and_values-array_api_strict-None-None]
SKIPPED decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='covariance_eigh',whiten=True)-check_array_api_input_and_values-array_api_strict-None-None]
SKIPPED linear_model/tests/test_ridge.py::test_ridge_array_api_compliance[Ridge(solver='svd')-check_array_api_input_and_values-array_api_strict-None-None]
SKIPPED decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='covariance_eigh')-check_array_api_get_precision-array_api_strict-None-None]
SKIPPED decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,svd_solver='covariance_eigh',whiten=True)-check_array_api_get_precision-array_api_strict-None-None]
SKIPPED linear_model/tests/test_ridge.py::test_ridge_array_api_compliance[Ridge(solver='svd')-check_array_api_attributes-array_api_strict-None-None]

# `test_array_api_train_test_split` inconsistency for Array API inputs. TODO cover by ticket.
SKIPPED model_selection/tests/test_split.py::test_array_api_train_test_split[True-None-array_api_strict-None-None]
SKIPPED model_selection/tests/test_split.py::test_array_api_train_test_split[True-stratify1-array_api_strict-None-None]
SKIPPED model_selection/tests/test_split.py::test_array_api_train_test_split[False-None-array_api_strict-None-None]

# Ridge regression. Array API functional support supported for all solvers. Not raising error for non-svd solvers.
SKIPPED linear_model/tests/test_ridge.py::test_array_api_error_and_warnings_for_solver_parameter[array_api_strict]

# PCA. Array API functionally supported for all factorizations. power_iteration_normalizer=["LU", "QR"]
SKIPPED decomposition/tests/test_pca.py::test_array_api_error_and_warnings_on_unsupported_params
# PCA. InvalidParameterError: The 'M' parameter of randomized_svd must be an instance of 'numpy.ndarray' or a sparse matrix.
SKIPPED decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_input_and_values-array_api_strict-None-None]
SKIPPED decomposition/tests/test_pca.py::test_pca_array_api_compliance[PCA(n_components=2,power_iteration_normalizer='QR',random_state=0,svd_solver='randomized')-check_array_api_get_precision-array_api_strict-None-None]

Documentation PR (if needed): #1340 (for example)

Benchmarks PR (if needed): IntelPython/scikit-learn_bench#155 (for example)

---

Checklist to comply with before moving PR from draft:

PR completeness and readability

• I have reviewed my changes thoroughly before submitting this pull request.
• 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 update 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 added a respective label(s) to PR if I have a permission for that.
• 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 summary table with measured data, if performance change is expected.
• I have provided justification why performance has changed or why changes are not expected.
• I have provided justification why quality metrics have changed or why changes are not expected.
• I have extended benchmarking suite and provided corresponding scikit-learn_bench PR if new measurable functionality was introduced in this PR.

[samir-nasibli]

Will be covered by ticket.

[icfaust]

Assume a follow-up ticket will be generated for private CI checks.