Fairness Through Regularization for Learning to Rank