MCODE: Fastest way to multiply

[Albert Chan]

Hi, Leviset

This is plain C code.

Although FFT is fast, it started to get inaccurate when numbers of digits go huge.
For product of over 2 million digits, it might return garbage results.

To be on the safe side, I cut that limit by half.
Div-and-Conquer code handled what FFT cannot.

These code are kinda heavy, and some application might not need the code bloat.
That was the reason for different versions, FFT_MUL = 0, 1, 2