softfloat: Use correct type in float64_to_uint64_round_to_zero()