Solarflare, it doesn't appear that we are making any headway with you on this issue. Perhaps it would help if we changed the scenario so that it basically happens in the reverse order.
Let's consider a totally different scenario:
We have a very long train, much longer than the length of the platform and so long we don't care where the front and rear are. But instead of just one observer, we have an observer at every window, all along the train.
Then on the ground, we have a platform of some arbitrary length. At the center of the platform, we have a flash bulb that will produce a bright flash of light sometime while the train is passing but we don't care when. Then at each end of the platform, we have a mirror placed at a 45 degree angle so that when the flash of light reaches it, it will reflect the light towards the train.
What we want to do is have whatever train observer is adjacent to the flash bulb when it flashes make note of the event. We'll call him Observer A. In a similar way, we want whichever two train observers see the light reflected off the two mirrors make note of those two events. We'll call the one toward the front of the train Observer F and the one toward the rear of the train Observer R.
Now isn't it obvious that by the time the light reaches the mirrors, the train has moved forward along with Observer A and so he will be closer to Observer F and farther from Observer R?
So now let's say that we take a video of this scenario and we play it backwards. We also pretend that Observer F and Observer R represent the ends of the train (except that now F and R are interchanged because the train will appear to be going backwards). Won't this be exactly the same scenario that we have been discussing in this thread, except that Observer A is not in the center of our now truncated train? And if Observer A sees the light from the two mirrors (which are analogous to the two lightning strikes) impenging on the flash bulb (analogous to the man in the original video) then Observer A cannot be the same as the woman on the train in the original video. In other words, if Observer A, who is not midway between Observer F and Observer R, sees the two flashes arriving simultaneously, then another observer who is midway between Observer F and Observer R cannot also see the two flashes arriving simultaneously.
Please note that this analysis does not require defining any frames of reference or synchronizing any clocks.
Does this make sense to you?