After confirming that the power is really off (a non-contact tester is helpful for that), I've always used a suitable diameter of carrot. You can usually find one that will fit into the bulb base. It is soft enough to cram in place, and woody enough to stay intact as you twist it. Don't eat the carrot after removing the base, unless eating light bulbs is something you enjoy doing...

+1 for routing to produce the extruded meeple. I suspect that just a couple of common bits combined with multiple passes are all it would take. Given the need to make more than two of them, that seems like the right way to approach the problem.

@mike, I suspect that most of the problem was with putting way more electronics without thought about the harsh environment they would live in. This was a niche brand device, and clearly suffered from lack of testing. Washer controls can get wet (and with just clean tap water in the right failure) and will collect wet lint. I've picked enough lint out of old-style mechanical timers... But quality whitegoods are built with the controls well isolated from the harsh environment and are likely more reliable today than what my mother experienced some fifteen years ago.

My mother suffered for years after buying a pricy imported washer/dryer single unit. Its CPU board was highly static sensitive. A clothes dryer might as well be a Van der Graf generator. It ate CPU boards all the way through its warranty, then it got to be too expensive to own quickly.

I've got a can on order to try some experiments. Reviews I've found online seem to be highly polarized, either "wow" or "ouch" and hardly any "meh". The bad reviews include a large percentage of obvious failures to follow directions, hence the urge to try it myself before buttoning the wall up.

This is in my home, and I'm aware of the small risks associated with glass mirrors. I also tend to believe (foolishly?) that my guests are not complete fools....

Thanks for the link to that calculator, I have other uses for that sort of information.

The wall finish is a plaster skim coat over drywall, in the "catfaced" style. The top coat is a skim coat applied with deliberate voids that are perhaps 1/32" deep and typically about a square inch of total area. I can paint a steel plate the same (or a deliberately contrasting) color, but I don't think I can readily match the texture. I'm not interested in machining a steel plate to have faux-plaster finish.

We have the room stripped down to studs at the moment, so floating plaster on expanded lath or putting something on either face of the drywall before final finish is certainly an option. This week. Soon, the options will be narrower.

@Reid, see my update. One pound flat object glued to three strong magnets. Does that seem plausible in your experience?