Add overlap warning to RATE

[erikcs]

It could be useful to add a warning and a suggestion to use the subset argument if RATE is called on data that appears to have low overlap (#1332), for ex:

Estimated treatment propensities take values between x and y and in particular get very close to 0 or 1. In this case, using the subset argument on forest$W.hat to filter data as in Crump, Hotz, Imbens, and Mitnik (Biometrika, 2009) may be helpful.