Machine ref to itself

[jonpsy]

#5139

Eren YEAGEEER!

@gf712 Some unit tests are failing still, submitting as early report.

[jonpsy]

It seems the failing tests haven't to do with my PR in particular, here's a report of failing test cases on develop branch and my branch

** develop **

	 54 - generated_cpp-evaluation-cross_validation_support_vector_machine (SEGFAULT)
	159 - generated_cpp-regression-random_forest_regression (SEGFAULT)
	253 - unit-MeanSquaredTest (Failed)
	565 - integration_meta_cpp-binary-domainadaptationsvm (Failed)
	575 - integration_meta_cpp-binary-svmlight (Failed)
	576 - integration_meta_cpp-binary-svmlight_batch_linadd (Failed)
	690 - integration_meta_cpp-pipeline-pipeline (Failed)
	695 - integration_meta_cpp-preprocessor-pca (Failed)

** machine branch **

	 159 - generated_cpp-regression-random_forest_regression (SEGFAULT)
        253 - unit-MeanSquaredTest (Failed)
        565 - integration_meta_cpp-binary-domainadaptationsvm (Failed)
        575 - integration_meta_cpp-binary-svmlight (Failed)
        576 - integration_meta_cpp-binary-svmlight_batch_linadd (Failed)
        690 - integration_meta_cpp-pipeline-pipeline (Failed)
        695 - integration_meta_cpp-preprocessor-pca (Failed)

This is true atleast in my local machine. This PR should be good to go, I'll let the members make the final call

[LiuYuHui]

I guess maybe there is some data race in cross_validation_support_vector_machine.

[jonpsy]

I guess maybe there is some data race in cross_validation_support_vector_machine.

#5121

[gf712]

this could be std::static_pointer_cast, since we know that this is a derived from Machine

[gf712]

Could this just be svm->train()->apply(features_test)->as<RegressionLabels>();?

[jonpsy]

Re-open when devs are back.