First, a trivial example that might anger you:

Let $A_i$ be the observables of the Mermin-Peres square, and $a_i$ their non-contextual values. Then $\prod_i A_i = -\mathbb{1}$, but $\prod_i a_i = 1$, contradiction. In this case $f$ is multiplicative. But the same contradiction can be obtained considering $\prod_i A_i+\prod_i A_i = -2\mathbb{1}$ and $\prod_i a_i+\prod_i a_i = 2$, where $f$ is neither multiplicative nor additive.

Now, a more interesting example, that I've found in a paper by Adán Cabello about inequalities for testing state-independent contextuality:

Let $$A = \begin{pmatrix}
Z \otimes \mathbb{1} & \mathbb{1} \otimes Z & Z \otimes Z \\
\mathbb{1} \otimes X & X \otimes \mathbb{1} & X \otimes X \\
Z \otimes X & X \otimes Z & Y \otimes Y
\end{pmatrix}$$

be the Mermim-Peres square. If one ascribes non-contextual values $a_{ij} = \pm 1$ to the observables $A_{ij}$, one can then prove that
$$ a_{11} a_{12} a_{13} + a_{21} a_{22} a_{23} + a_{31} a_{32} a_{33} \\+ a_{11} a_{21} a_{31} + a_{12} a_{22} a_{32} - a_{13} a_{23} a_{33} \le 4, $$
whereas in quantum mechanics
$$ \langle A_{11} A_{12} A_{13}\rangle + \langle A_{21} A_{22} A_{23}\rangle + \langle A_{31} A_{32} A_{33}\rangle \\+ \langle A_{11} A_{21} A_{31}\rangle + \langle A_{12} A_{22} A_{32}\rangle - \langle A_{13} A_{23} A_{33}\rangle = 6.$$
The proof of the inequality may be done simply by enumerating the $2^9$ possibilities, if you're lazy, or by playing around with the triangle inequality. In either case, we have an $f$ that's not additive nor multiplicative. Of course, in this case the contradiction takes the form of an inequality, instead of a definite value for non-contextual values.

I guess then that they used always a multiplicative or additive $f$ because it's easier to construct these kind of contradictions, based on parity arguments. But I don't think there's anything fundamental to it.

This post has been migrated from (A51.SE)