Introduce arbitrary-precision floating-point type Decimal
Since we're dealing with sensitive data in general with VF, it's a wise choice to obsolete IEEE 754 (which is used for inexact calculations) in favor of IEEE 854-1987 (which is used for exact calculations).