MI-SMM, MI-SVM methods create objects with different components

[Sean Kent]

Need to unify the list components of these objects (as well as possible, ok to have some differences), and add a method parameter into the constructor. The method parameter gets used in the "predict" function to create a unified set of outputs.

Some initial steps to think about

• Build mildsvm object first with the mild object in mind
• Write predict.mildsvm with predict.mild in mind
• Adjust everything that makes a mild object to be a mildsvm object
• Incorporate the mild predict into mildsvm predict

[Sean Kent]

Note: a 'mildsvm' object that is fit with method = 'heuristic' contains two copies of the training data, when only one should be needed