ENH: standardized coefficients - params table

Add beta coefficients (coefficients for standardized data) as a post-estimation results method.

API: return similar to t_test with params table, summary, summary_frame

Most likely this can internally reuse t_test, we just need the linear transformation matrix to standardize. Alternative would be similar to transformation in models, e.g. fit_constrained and returning a results instance.

Applies to linear model, but I never looked at the interpretation in GLM, discrete. So, I'm not sure whether it would be helpful to add in base instead of regression.

Issue #1131 discusses transformation to improve estimation but here it is just a post-estimation diagnostic.

I wrote this function already a long time ago, but never committed it anywhere. If the design in not orthogonal, then the beta coefficients are not a variance decomposition. I read some papers for importance statistics in that case, but those didn't look very popular.

related topic: SAS has a large number of extra statistics and diagnostics for the linear model.