Materialize train test

[HoustonJ2013]

Closes #470.

Add the col_stats in the dataset materialize function, allowing the column stats calculated from train dataset to be used in the test data materialization.

train_df = ..
test_df = ..
train_dataset = Dataset(train_df, 
                 col_to_stype=col_to_stype, 
                 target_col="target")
train_dataset.materialize()
test_dataset = Dataset(test_df, 
                 col_to_stype=col_to_stype, 
                 target_col="target")
test_dataset.materialize(col_stats=train_dataset.col_stats)

[weihua916]

Thanks! this makes sense. can we do the following?

• Add basic validation logic and raise error if col stats is not appropriate.
• Add a unit test.