The board I have is identical to the one in the first two pictures in the OP of this thread. For testing, I'm using an XML, powered by a single Samsung ICR18650-28A laptop pull. It was at 4.2v at the start of testing. Using fairly thin long jumper cables, going through an Omten clicky. I was initially using short 22awg, but changing to the jumper cables/switch made no difference either way to the problem.
This morning, I re-soldered each component on the back of the board (CAP, Diode, & the two resistors), one at a time, & tested each time, with the problem persisting. I also tried the tweezers bridging over the diode.
I previously mentioned that mode memory didn't appear to work - to clarify that, it had 'next mode' memory.
I bridged star 4 to ground, then it would only operate in medium mode - nothing else, no pulses, flashes, or mode changes.
Not really getting anywhere, I decided to move the CAP back to C1, & replace the bridge on R3 with the 100 ohm resistor, exactly like in the first two pictures in the OP of this thread.
Problem is now gone...
I would have liked to get a more definitive result using the 'v2' cap arrangement, but this driver is going into a light for a workmate, & right now it's working, allbeit with the spikes that only me & you know about....
I'll put together another board, using the same layout I had the problems with, just to see if the same thing happens.