It seems to me that most arguments in favor of impossibility of communication from beyond black hole horizon region are based on "test-particle" scenario, where the falling object is (very) light with respect to the black hole mass. Argument is as follows: everything, including gravitation, propagates at speed $v \le c$, so it is within future light-cone which (the light-cone) ends-up in singularity. QED.

Does, without exception, this argument hold also for very massive falling objects which importantly contribute by themselves to space-time modification? Imagine a neutron star which is just under the horizon divided into two parts (in non-gravitational way, using e.g. atomic explosion), which then fall separately onto singularity on different trajectories. In alternative scenario the neutron star is not divided....Are alternatives completely hidden to the external observer?

Let me forward a simplistic argument: Imagine the neutron star falls direct on the singularity. In non-divided scenario the neutron star goes quickly away from the horizon and produces no "polar angle" effects on space-time. In "divided" scenario two parts can (in extreme case) make many rotations (so very long time) just under the horizon (almost speed $c$) slowly falling. By continuity argument (functions are continuous) I would expect an effect on the horizon which is in their very proximity. And the horizon is an observable to the external viewer... Non-negligible masses do contribute during the fall to the horizon shape - do not they?