As mentioned, the actual zener print is "4B7", so the idea they used a lower value on purpose does not hold. 4B7 seems a very plausible print for a 4V7 zener.
Tested first with a component tester (unknown test current) and then to confirm with a 22K and then a 10K resistor on about 8V power supply. All 3 gave about 3.3V. Also used tried two different multimeters, just to make sure. Weird.
I now have all active components (FET, chip, diodes) in that part of the circuit replaced. Trafo has continuity. But 340VDC goes in (I checked the DC voltage after the full wave rectifier), nothing comes out. Guess I have to start measuring if I get any waveforms on the chip. Not sure I like hooking my scope to the hot side of the board as I have no diff-probe. (and yes, I know about the ground loop trap)
Anyway, about the zener. Still unsure how it is possible that a 4.7V zener fails to become a seemingly good working 3.3V zener. But assuming the manufacturer accidentally mislabeled a batch of zeners seems equally unlikely.