Reenigne,
I just looked at the schematic for the 5151 for the first time, it appears that the horizontal output stage is driven directly from H sync, there is no local horizontal oscillator stage, which is a disaster waiting to happen in my opinion. If the horizontal output transistor is driven into conduction too long by an abnormal H sync, then more energy is stored at the end of scan time and the flyback voltage likely will punch through the H output transistor, although they have provided some protection. This design technique seems to be unique to the world of computer monitors. Even from the late 1930's all TV sets had their own independent H oscillator. Generally these gave little trouble, even with line by line sync pulses, because the sync pulse gets differentiated for its leading edge (is coupled into the oscillator with a very low value capacitance) and couldn't cause the output device to stay on too long by excessively modifying the drive waveform out of the oscillator. I don't know why IBM decided to make a monitor with no self contained horizontal oscillator. Maybe they considered the H sync was so stable and clean, being digitally generated (rather than out of a TV sync separator) it would be ok, but it saved so few parts it wasn't work the risk.
If I owned a 5151, I would put a small H oscillator board in it, and H sync it per line with a small value coupling capacitor from the H sync pulse feed. Free- running R-C oscillators are very easy to synchronize with a small pulse injected just before they are due to change state.
PS: luckily the amber monitor in the 5155 computer has its own H oscillator so when its fed a 21.85kHz EGA image and loses H & V lock and the picture "scrambles" no harm comes to it.