Glance over at the
Scripting Database topic in the DROD Architecture forum, and you'll find that the scripting engine is powerful enough to, say, code a reasonable pseudo-RNG, which is an example of an algorithm that definitely deals with large numbers and overflow. We don't have the option of unsigned integers, but we also don't have the option of toggling overflow for when it might actually be useful.
Workarounds exist for both, but I find trying to script overflow modulus in an environment where overflow either errors or clamps the value to be quite a bit more taxing than scripting clamping in an environment where overflow is allowed. You can script your own clamping against overflow for multiplication, addition and subtraction in just a few lines (though it'd be a few more if you don't know the sign of both values you're multiplying, but still easier than scripting overflow in an environment where you're not allowed to.)
In any case, I'd hope an architect would be careful when using such large numbers whatever the reason anyways, whether overflow is a possibility or not.