Thanks for being patient.
Tried both the Darlington pair and, after rearranging the circuit a little bit, the Szilaki pair. Essentially same result with both. The circuit becomes far too sensitive to even trivial amounts of ambient light. The installation area is generally dark but not pitch black. The ambient light level varies somewhat.
Tried series R up to 10M between diode and base. No joy. I then tried a divider network on the base. This worked but by the time I biased it down enough to solve the room light problem with sufficient margin, my object / no object voltage swing was quite narrow, less than 1v. No good.
All this lead me to ask why boost the signal if I just turn around and clamp it down? Must be the tea cup thing. That made me wonder just how much of a jackhammer my tea cut is resting on. I rebuilt the original single transistor circuit and incorporated your suggestions except the clamping diodes. I ordered some 1N5817. I used 30cm of shielded cable between diode and Q base. I tried every device I could find - cell phone, cordless phone, CFL, electric drill, cordless drill, AC grinder, even an old B&D router that throws so many sparks from its brushes it should be in a 4th of July parade. Nothing. No false triggers. The circuit carried on working seemingly unfazed by any of the devices operating within inches of the cable.
Is it possible the shielded cable is sufficient? It sure seems to be. Should I throw a couple clamping diodes on it when they arrive and call it done?