The keycaps on the handwired boards have started to split between layers. If that layer happens to be at the stem, it means the entire top of the keycap comes off (leaving a stem stuck in the switch that cannot be extracted). Spending ages tweaking printer settings trying to get a stronger keycap seems futile and those of you who have been waiting so long shouldn't need to wait ages for the CNC mill to arrive, so I've had a close friend help me get some aluminium keycaps going on a borrowed mill.
In the meantime, I have been testing square relegendable keycaps on my Planck to see if they are more comfortable and usable than trying to type on standard keycaps. If so, this might be an alternative way of solving the keycap difficulties.
I've also tested the switches now that the oscilloscope has arrived. There seems to be no problem on the hardware side, surprisingly. There must be some quirk in the debounce code that I missed, so I'll be rechecking that as well.