Add regularization

[hammer]

scikit-learn supports the following penalties:

• [x] penalty="l2"
• [ ] penalty="l1"
• [ ] penalty="elasticnet": Convex combination of L2 and L1; (1 - l1_ratio) * L2 + l1_ratio * L1.

[hammer]

l2 implemented with https://github.com/hammer/olearn/commit/c6276abad2035e889e8f2e2ce12d9f77e1040e83