Consider adding a new Fitted() generic

[bgreenwell]

Idea is to make it easier for users to construct their own fitted-vs-residual plots. It's capitalizes to it does not interfere with R's built-in generic and the specific packages that have methods defined.

For example:

Fitted.polr <- function(object) {
  object$lp - object$zeta[1L]  # Xb - a1
}