reuse residuals of x

[laszlosandor]

If one is fine using xq, why not allow using the saving and reusing x_r from before? It is always "risky" as a changing sample could render the values invalid or something, but the issue is similar with xq. On the other hand, when is plotting multiple outcomes against the same x, it is pointless to redo the _regress and the _predict all the time.