Use Math.ulp in FloatingDecimal

The code in sun.misc.FloatingDecimal has logic for computing the ulp, unit in the last place, of a floating-point value. This is redundant with the library math Math.ulp; the library method's logic should be used instead.