Say it quietly, but... I think we have a win.
In between visiting scrapyards and the like this morning, I visited one of my stores to grab some 120kΩ resistors to calibrate my Leader CRT tester (next on the bench - I didn't like the burnt resistors in the PSU, nor the split sleeve on one electrolytic) and also some 270kΩ to try in the PSU. This also gave me an excuse to exercise the 'scope and its 100x probe.
So... connect the 'scope to the collector of the chopper transistor and wind up the variac. These traces are taken with the o/c 270kΩ resistor still in place. This is with the PSU running at about 200V or so AC.
Having disconnected, and allowed the two smoothing caps to discharge, I replaced the o/c 270kΩ resistor with a new one and again, wound the variac up to the same voltage. In all this time, the 'scope has not been touched.
Finally, the moment of truth: will it take a cold start?
The answer is.... yes! There is a heart-stopping moment though because the longest second or so ever passes between switching on and the PSU bursting into life! So finally, I've ordered some 350V 1W 270kΩ resistors to go in the position previously occupied by the o/c one as I'm unsure of the rating of the temporarily fitted one.
I'm left wondering if the cold start produced a spike that exceeded the SOA of the switching transistor when that 270kΩ was open circuit and punched a hole in the junction...