`SVC`/`SVR` `kernel` and `degree` are never passed to `scikit-learn`

[lars-reimann]

Describe the bug

This issue affects SupportVectorMachineClassifier and SupportVectorMachineRegressor:

These models never set the kernel of the scikit-learn estimators when fitting. Thus, the default value ("rbf") is always used. Likewise, the degree is not set if a polynomial kernel is chosen.

To Reproduce

Check the code of their _get_sklearn_xy methods.

Expected behavior

The selected kernel should actually be used.

Screenshots (optional)

No response

Additional Context (optional)

No response