Non-coinciding event horizon and apparent horizon 
Proposition: the event horizon and the apparent horizon of a black hole always coincide.

As a reminder: the event horizon is defined as the boundary of the closure of the causal past of future null infinity, i.e. the boundary of the closure of the set of points which can reach infinity by a timelike path (so the boundary of the black hole). The apparent horizon is defined as the boundary of the set of trapped surfaces, where a trapped surface is defined as a surface for which all outgoing null rays converge.
Question: The above proposition is known to be false. Can someone give a counterexample? As always for counterexamples, the simpler the better!
I know the following:


*

*For a counterexample we necessarily need to consider a non-stationary black hole, as Hawking proved that the proposition is true for stationary ones.

*The apparent horizon is always inside the black hole. But doesn't this imply that there are points inside the black hole from which light rays can diverge?

*Apparently, one counterexample is given by stellar collapse, though I fail to understand this counterexample, i.e. I don't recognize the points which are outside the apparent horizon and inside the black hole.

*Unlike the event horizon, the apparent horizon is observer dependent. Wikipedia states: "For example, it is possible to slice the Schwarzschild geometry in such a way that there is no apparent horizon, ever, despite the fact that there is certainly an event horizon." Clearly they don't coincide then, but doesn't this contradict the fact that they do coincide for stationary spacetimes?


So basically, I know some differences between the two concepts, but I cannot think of a concrete example for which we can identify the apparent horizon and the event horizon, and see that they are different.
 A: One simple counterexample is the observer moving in the Minkowski spacetime with constant acceleration. This observer perceives an apparent horizon - Rindler horizon, while there is no event horizon at all (and therefore no black hole). The existence  of this apparent horizon leads to the Unruh effect - black-body radiation measured by such observer.
A: Consider a collapsing shell of dust in what is otherwise flat spacetime. Outside is Schwarzschild spacetime, inside it is Minkowski spacetime. There is a moment in time when the shell reaches its Schwarzschild radius $r = 2M$. The event horizon forms earlier, at the centre, and expands outward at the speed of light to reach $r = 2M$ at the same time the shell does. After that, the event horizon remains at $r = 2M$ forever.
The apparent horizon forms when the shell crosses $r = 2M$, and splits with one apparent horizon remaining at $r = 2M$, and another apparent horizon following the shell down to $r \rightarrow 0$. So the latter does not correspond with an event horizon.

This example is taken from an excellent pedagogical paper by Matt Visser, "On the observability of horizons" (2014), see his Figure 1.
A: The Vidya metric is a counterexample. Any non-stationary horizon is as well. 
$$ds^2 = -f dv^2 + 2 dv dr + r^2 d\Omega^2$$
Sections 5.1.7 and 5.1.8, starting p. 171 of Poisson's "Relativist's Toolkit". 
