math.factorial() was accepting non-integral Decimal instances. This is inconsistent with the actual behaviour for floats, which are not accepted.