[PATCH] [ConstantRange] Add a smultiply method for signed multiply

[Sanjoy Das]

> clearly identified a deficiency. ConstantRange::multiply could compute it
> twice once with zext and unsigned min/max and again with sext and signed
> min/max, then choose the one with the smaller range. Maybe there's an even
> better algorithm.

Do you think there is value in being more precise if one of the ranges
has one element (or, by extension, a small number of elements)?
Finding a precise range for [a, b) * { c } may be a much easier
problem to solve than finding a precise range for [a, b) * [c, d) if
the former shows up frequently enough.

-- Sanjoy