normalize preprocess for the future use of pipeline

[Qi-max]

For the future use of pipeline, the classes in matbench.preprocess can also take the form of fit + transform, as the classes insklearn.preprocessing, such as Imputer and LabelEncoder etc.

[ardunn]

If this is the case, similarly, AutoSKLearn class should implement .predict and .fit, yes?

[ardunn]

I don't think the overall pipeline of matbench will take the form of an sklearn pipeline, since most of our underlying code uses dataframes and sklearn.Pipeline doesn't play nice with DataFrames sometimes...

@Qi-max can we close this issue?

[ardunn]

Addressed more broadly in #92