Issue Description:

pow(_ x: Decimal, _ y: Int) is a thin wrapper around NSDecimalPower. NSDecimalPower`s power argument is an NSUInteger but Swift exposes it as an Int, which gets casted as a UInt (presumably so the code can be translated in a straightforwards way from Foundation.framework). This means that the "reasonable-looking" pow(10 as Decimal, -2) is really "pow(10 as Decimal, 18446744073709551614 as UInt)". Internally this causes an overflow which is then exposed as a NaN because it seems like Decimal doesn't actually have a way of representing infinity. In the absence of the latter being fixed, it would be nice if the function at least took a UInt or had behavior that worked for negative integers.