[docs] [issue14245] float rounding examples in FAQ are outdated

[John Roth]

I keep looking at this issue. What rattles around in my head is the 
satori I had when I first heard "integers are for things that are 
counted, floats are for things that are measured." Possibly something to 
the effect that "the industry standard for floating point is intended 
for measurements, which are never absolutely precise. Thus floating 
point, by its nature, will not usually represent factions as precisely 
as one would like."