Various IEEE floating-point number constants

`unitRoundoff : Double`

Largest number that can be added to a floating-point number without changing

its value, i.e. `1.0 + unitRoundoff == 1.0`.`epsilon : Double`

Machine epsilon is the smallest floating-point number that distinguishes two

floating-point numbers; the step size on the floating-point number line.`nan : Double`

Not a number, e.g. `0.0 / 0.0`. Never equal to anything, including itself.

`inf : Double`

Positive Infinity. Can be negated to obtain Negative Infinity.