Integrate the secSVM classifier made in base python

We should add the base python implementation of the secure SVM classifier from "Yes, Machine Learning Can Be More Secure! A Case Study on Android Malware Detection". See snippet for draft implementation: $24

This should be put in the new package: ml.classifiers.secure.

Include unittests to compare the secsvm to a standard svm, by checking the alignment of the decision boundary. Add a 2D and a higher-dimensional example.

Related to #184 (PyTorch implementation)