I guess you got me wrong. I am not saying that distinct values are alone of interest for optimizer. I am trying to say that distinct values along with the total number of values in that range (which qualifies) is much more useful than RCD alone. I know as of now there are no distinct values in histogram because it is not being captured so it has RCD as the only its option.
Regarding each column with stat. I mentioned it wrong , I wanted to say each column with SARG or leading column in index or in some of cases non leading column of indexes.
Thanks