Get everything together for an official beta release, with much more stable naming, documentation, and interface. Short-term targets include:

• Vignettes on 1) basic use, 2) distributional estimation, 3) changes-in-changes model
• Consistent naming conventions throughout the package
• Maybe replace quantreg fortran binary for sfn.start.val (if possible)

Ideally we would also include:

• Delta method standard errors where possible
• Helpers w/ the mice package for multiple interpolation
• Our own cross-validation routines for lasso, since the ones from rqPen don't seem to pass methods correctly
• add quantile random forests model, since this could be breaking (better to break pre-beta perhaps)