QuantTensor's __truediv__ always results in a NaN zero-point when both inputs have a 0 zero point

[nickfraser]

Spotted while working on #740. The following branch:

            if self.is_zero_zero_point(self) and self.is_zero_zero_point(other):
                output_zero_point = self.zero_point / other.zero_point

Ensures that the output always has a NaN zero point (i.e., 0 / 0 = NaN). I assume we would like the output zero point to also be zero.

[nickfraser]

Closed by #768.