When converting a double larger or equal to math.MaxInt64 to an integer, it previously overflowed silently. It now returns an explicit error.
This fixes along the way, the math.abs(-9223372036854775808) issue.
* Add math.acos and math.abs
* Add acosh, asin, asinh, atan and atan2
* Test for proper function documentation
* Cast to double on math functions
* Use the new types.Value interface