In On $l$-independence of Algebraic Monodromy Groups in Compatible Systems of Representations, they stated such an example (Example 6.3):
Let $X$ be a connected normal scheme of finite type over $\text{Spec}(\mathbb{Z})$ and of dimension $\geq 1$. Let $K$ be the function field of $X$ and $\mathcal{G}=\text{Gal}(K^s/K)$. For every closed point $x$, let $K_x$ be the function field of the henselization of $X$ in $x$, and $K_{\bar{x}}$ that of the strict hensalization. We have canonical embeddings $K\hookrightarrow K_x\hookrightarrow K_{\bar{x}}$. Every extension $K^s\hookrightarrow K_x^s=K_{\bar{x}}^s$ of the embedding induces a homomorphism $j: \text{Gal}(K_x^s/K_x)\rightarrow \text{Gal}(K^s/K)$. Let $A$ be the set of all triples $\alpha=(x, j, F)$, where $F$ lies in the $j$-image of any representatives of Frobenius in $\text{Gal}(K_{\bar{x}}/K_x)$. By the density of Frobenius conjugacy classes, we have an $F$-group.
I could understand everything except the last sentence: what is the theory about the density of Frobenius conjugacy class? Is there a Chebotarev density theorem for a function field of higher transcendental degree? I did not find this on Internet, and appreciate every help or explanation! Thanks.
