update the stepwise learning function.

[kan-qi]

The current implementation of the stepwise regression model:

The stepwise_linear_model initiate the model for some hyperparameter.

m_fit.step_lnr and m_predit.step_lnr will be alternatively called during the bootstrapping and cross validation processes.

try run the model training and testing processes from this file:

• un-comment the step_lnr function..and comment out other models

And then run this benchmark function:

The wikipedia page provides some background information. https://en.wikipedia.org/wiki/Stepwise_regression