Implement faster multiplication algorithm

[isaacholt100]

Currently, the multiplication algorithm for integers is long multiplication, which is slow for large numbers (O(n^2)). For integers of size beyond a given threshold, faster algorithms should be used such as Karatsuba, Toom-Cook, and maybe also Schonage-Strassen. This will also allow a divide and conquer division to be used which will have the same time complexity as the multiplication algorithm.

[krakow10]

Can this be done at compile time based on the digit count or will that require const generics?

[isaacholt100]

Hi @krakow10, sorry this comment completely slipped my attention... yeah we could use a conditional statement based on the digit count which we would evaluate at compile time. I hope to make good few changes to bnum for the next version (e.g. include initial support for arbitrary size floats), and I'll try to add at least Karatsuba in to these changes.