Add RF clsf inference hyperparameters

[ahuber21]

Description

Adds hyperparameters to random forest classification. Adds all the necessary glue logic in oneDAL to set the parameters in the DAAL kernel, to which the oneDAL implementation falls back.

Changes proposed in this pull request:

• Replace hard-coded #define statements in df_classification_predict_dense_default_batch_impl.i with oneDALs hyperparameter framework
• Per default, the non-AVX512 values are configured, AVX512 will be configured through the Python interface (to be added)

[ahuber21]

/intelci: run

[ahuber21]

/intelci: run

[Alexandr-Solovev]

/intelci: run

[ahuber21]

/intelci: run

[ahuber21]

I think the Pre-commit step shouldn't have failed, restarted it here http://intel-ci.intel.com/eee09227-f812-f180-afe8-a4bf010d0e2e

[ahuber21]

The combination of the two runs shows all green checks.

[ahuber21]

@Vika-F I have addressed your first two points, let's hope the CI is still green. For the third point, I will create separate tasks!

[ahuber21]

/intelci: run

[ethanglaser]

/intelci: run

[ethanglaser]

CI is in good shape. @Vika-F if changes look good feel free to merge it

[Vika-F]

@ethanglaser Yes, 90+% of the work is done here, need to implement the test and add a couple of changes. I've moved the work into https://github.com/oneapi-src/oneDAL/pull/2692. Will merge that one as soon as it is ready.

[ethanglaser]

@Vika-F can this one be closed now?

[Vika-F]

Closing this PR as https://github.com/oneapi-src/oneDAL/pull/2692 was already merged.