Alright, folks, I took Amberwolf's suggestions to heart, and I thought I'd pass along what I figured out.
I had been using a conventional 12-pole PAS sensor like this:
https://bmsbattery.com/parts/705-dual-hall-sensor-12-signals-easy-assembling-pas-ebike-kit.html
With an "external" bottom bracket (BBR60), there's nothing even remotely small enough in diameter to mount a split PAS sensor around, even if I want to dremel out all but the thinnest shell.
I ended up sticking 5 N52 (strong!) neodymium 12mm x 2mm magnets onto the five chainring bolts and zip-tying the hall sensor to the frame. Note: this will only work for steel chainring bolts; some are aluminum.
I don't know if north or south is pointing towards the hall sensor, but I had to place the magnets in the same orientation as the originals were. This is very obvious: one way will stick to the magnets in the original assembly and one one way, and one way works and one way doesn't.
At first, the motor "surged" on and off as I rotated the pedals. I went back and discovered that all but one of the magnets were oriented correctly; (I tested with another magnet to check for repulsion/attraction). Once they were flipped around the motor ran smoothly down at a 45-50 rpm pedaling cadence.
The sensor needs to be oriented so that the magnet runs along the length of the sensor, not just merely across it (which at first looks contrary to how bicycle computer pickups are mounted).
This should get me down the road for a while, but I don't expect it to be a longterm solution, especially the zipties to hold the bracket on. I don't fancy the idea of that wired sensor getting caught up in the cranks and chain.
Also, another improvement would be to mount 10-12 magnets onto the chainring itself, perhaps with superglue or 5-minute epoxy.
Hope this helps somebody.