Extend mod to complex numbers properly

[gwhitney]

Describe the bug
The mod function has a reasonable definition for complex numbers: for any complex numbers w and nonzero z, there are at most four Gaussian integers q (complex numbers with integer coefficients) such that w = zq + r, where norm(r) < norm(z). Typically, we can choose the unique r of minimum norm, and define mod(w,z) to be this r (and the corresponding q to be the quotient). Occasionally two or four of the possible r values tie for smallest norm. In such a case, we can break the tie by taking the one that has the corresponding q of smallest norm.

To Reproduce
math.mod(math.complex(1,4), math.complex(1,1))

Currently this throws a TypeError; it should return i because 1 + 4i = (2+i)(1+i) + i. Note that Wolfram Alpha returns 1 because it is also the case that 1 + 4i = (2+2i)(1+i) + 1. However, the possible quotient 2+i has smaller norm than 2+2i, so it is preferable to choose that quotient and return remainder i.

[josdejong]

👍 Anyone interested in implementing support for complex numbers in function mod? Help would be welcome.

[gwhitney]

Oh, I will surely take care of this in the sequence of PRs to implement #3374; that's where the point came from, really.