[Python-ideas] Re: math.copysign not to introduce float