The version of the mean value theorem you are using establishes the existence of $c_1 \in (y,y+\eta)$ such that
$$ \frac{u(x+\xi, y+\eta) - u(x+\xi,y)}{\xi + \mathrm{i}\eta} = \frac{\eta}{\xi + \mathrm{i}\eta}u_y(x+\xi, c_1) \text{.} $$
This is not the form given. To get $c_1 = y + \theta_1 \eta$ we must know that $u_y(x+\xi, y + \theta_1 \eta)$ takes every value in the set
$$ \left\{ u_y(x+\xi, c) \mid c \in (y,y+\eta) \right\} \text{.} $$
An easy way to do so is to apply the intermediate value theorem, which here requires that $u_y$ is continuous in its second argument. That is, $u$ is continuously differentiable with respect to $y$.
The other term in your long equation give that $u$ is continuously differentiable with respect to $x$. The same development for $v$ gives the requirement that it is continuously differentiable in each of $x$ and $y$.
One hypothesis that gives all four of these is that $f_x = u_x + \mathrm{i} v_x$ and $f_y = u_y + \mathrm{i} v_y$ are continuous, so $f$ is continuously differentiable with respect to both $x$ and $y$.
Maybe a weaker hypothesis would work. $f$ merely differentiable in $x$ and $y$ is insufficient because there is no analog of the intermediate value theorem that can give the $\theta_1$ and $\theta_2$ for potentially nowhere continuous derivatives. See this answer for a reminder of how awful the derivative of a differentiable but not continuous differentiable function can be -- and an easy visual example of why $\theta_1$ (and $\theta_2$) need not exist for such a function.