First let me give some cheap examples of CY manifolds that satisfy condition of 1) while they are not tori quotients.
Example. Let $X$ be any CY manifold with $h^{1,1}(X)=1$, for example $X$ can be a quintic in $\mathbb CP^4$. Then the conclusion of 1) holds, because all Kahler Ricci flat metric on $X$ are proportional. Now, to get an example with $h^{1,1}(X)=2$ one can just take a direct product $X_1\times X_2$ of CY manifolds satisfying $h^{1,1}(X_i)=1$.
So the list of manifolds with $c_1=0$ satisfying 2) is larger than tori quotients. But maybe all these examples can be understood. For example, it should be easy to show that a $K3$ surface does not satisfy 1). I'll give a brief proof in the case of a generic K3
that does not have $-2$ curves.
EDDITED. I put more details here so that it becomes clear that was said previously is correct.
The good thing about harmonic forms is that the sum of two harmonic forms is harmonic.
By Nakai-Moishenzon (for Kahler surfaces), the Kahler cone
of a K3 without $-2$ curves coincides with a connected component
of $(1,1)$ classes with positive square in $H^{1,1}$.
Now, by Yau in any such class there is a unique Ricci-flat metric. And also obviously
there exists a unique harmonic metric. Suppose that assumption 1) holds for our K3.
Then it is obvious that this means that each harmonic form $w$ with $\int_{K3}w^2>0$
in the correct component of the cone
is Ricci flat. Let me get a contradiction from this
So let $w_1$ and $w_2$ be two Kahler Ricci-flat forms harmonic with respect to the metric defined by $w$ on a $K3$ surface, and moreover that $aw_1+bw_2$ is Ricci flat again. Then we know that $(aw_1+aw_2)^2$ is proportional to $\Omega\wedge \bar\Omega$, where $\Omega$ is the complex volume form on $K3$. Consider now the family of forms $w_1-tw_2$, $t>0$. Take the first moment $t_0$ when we have $\int_{K3}(w_1-t_0w_2)^2=0$. But since $(w_1-tw_2)^2=c_t\Omega \wedge \bar\Omega$ for some $c_t>0$ for all $t$ less than $t_0$ (since this form is Ricci flat)
we conclude that $(w_1-t_0w_2)^2$ is equal to zero point-wise. So the kernel of $w_1-tw_2$ should define a holomorphic folitation on $K3$. Since for $K3$ it holds $h^{1,1}=20$, we should get a tremendous amount of holomorphic foliations on it, but this is clearly impossible.
This is still not a 100% complete reasoning, but I am sure it can be completed. So it seems to me that the complete list of manifolds satisfying 1) are all manifolds that are finite quorients of a torus times a collection of CY manfiolds with $h^{1,1}=1$. The idea is simple: whenever you have two Ricci-flat forms such that $w_1+tw_2$ is Ricci flat for small $t$, this should lead to a local metric splitting of the manifold into direct product. Then we should just use De-Rahm decomposition theorem.
In his paper Invariant Kahler metrics and projective embeddings of the flag manifold, Bull. Austal. Math. Soc. 49 (1994), K. Yang considers the flag manifold $$F_{1,2,3}(\mathbb{C}^3):=SU(3)/S(U(1)^3)$$
and determines the space of invariant Hermitian and Kahler metrics on it.
In particular, he shows that a Killing metric is not Kahler.
On the other hand, by applying Kodaira embedding theorem, he proves that $F_{1,2,3}(\mathbb{C}^3)$ is projective algebraic and provides an explicit projective embedding of it.
The computations are made quite explicitly in terms of the Maurer-Cartan form of $SU(3)$.
So this could be one of the examples you are looking for.
Best Answer
You don't have to use hard Lefschetz to conclude $df=0$ from $\omega\wedge df=0$.
This is a linear algebra fact, valid pointwise : if $\alpha \in T_x^*X$ satisfies $\omega_x \wedge \alpha=0$, then $\alpha=0$ (of course, assuming $\dim_R X \geq 4$.
The short argument is that, $\omega_x^{n-1}\wedge : T^*_x X\to \bigwedge^{2n-1} T^*_x X$ is an isomorphism ("pointwise not so hard Lefschetz", so to speak).
This said, as in Francesco's answer, you can have non proportional conformal riemannian metrics that are Kähler with respect to different complex structures, so that the corresponding 2-forms are no longer (pointwise) proportional.