I suspected a CDI coil wouldn't put out enough voltage for the "energy transfer" -- the worst misnomer ever -- system and suggested shorting out the spark plug electrode on one of the 2 cylinders by pounding the gap shut or stuffing it with a bit of foil.
The theory was instead of jumping two gaps it would only need to jump one. The lower effective resistance would give a better spark on the remaining firing cylinder. The al foil trick worked. It ran on one cylinder which seems to confirm my suspicions about the coil as well as my test, which may or may not be original. Googling is hard work.
Anyway the engine has two magnetos opposite each other, one for the ignition and one for charging the battery. The four magnets are equally spaced on the flywheel.
The 2 low voltage magneto coils look about the same and both seem to do the same thing to a volt meter when cranked -- admittedly not a very scientific indicator.
It seems like it would be possible to just forget about charging a battery and wire another CDI coil to the charger coil.
Even simpler would be to tap the plug gaps to half that recommended in the specs.
Maybe the fuel efficiency would drop by 20%? Maybe it wouldn't fire at all?