To be fair functions like `abs(x)` are expected to return int when x is int and float when x is float. Factorial acts same with respect to types so it is not th

To be fair functions like `abs(x)` are expected to return int when x is int and float when x is float. Factorial acts same with respect to types so it is not that unreasonable. The real problem is that `fact` is not the actual factorial but its approximation (doubly so because it goes through float anyway). So all extra precision provided by bigint is useless.