On the thread cited I gave a three state, two outcome counterexample to P2 which does just that. Having two outcomes obviously a utility function is not an issue. (It can be extended it with an arbitrary number of "fair coins" for example to satisfy P6, which covers your actual frequency requirement here)

My weak claim is that it is not vulnerable to "Dutch-book-type" arguments. My strong claim is that this behaviour is reasonable, even rational. The strong claim is being disputed on that thread. And of course we haven't agreed on any prior definition of reasonable or rational. But nobody has attempted to Dutch book me, and the weak claim is all that is needed to contradict your claim here.