Extend `mod` to complex numbers properly

**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`.