Remembering back to my old TV repair days when the 2nd anode, due to a Flyback transformer failure, would do the Tesla thing, corona glow and all.....Arcing like mad until the x-ray circuit kicked in (if it kicked in). Only got bit once over those years, 40KV must have been..............should of stopped then but, noooooooooo
Thinking it's either the stator (exciting coil) and/or the CDI (Switch Box).
Basically, the CDI Ig. uses the induced voltage (magneto if you will) generated from the flywheel's perm magnets passing by the specific exciter coil's in the stator to charge a capacitor (some use an internal oscillator tanking at some resonate LC freq, whining away). that lives in the CDI box (Switch Box)..The "C" (capacitor) in the CDI box is in series the the primary side winding of the ig. coil.
To fire the plug, the charged up cap is discharged (SCR (or FET) grounds the cap via signal from the "other" flywheel stator trigger winding) through the ig. primary coil winding and through the magic of Faraday's Law, the "turns ratio" creates a proportional stepped up voltage on the secondary winding of the ig. coil of say 20KV that arc's at the electrode of the plug.
Say something like 1:100, so if the cap normally charges to say 200VDC on it the secondary of the ig.s coil would have 20,000VDC
If that capacitor becomes over charged somehow by a fault/s in the stator 's (shorting winding's and the like) exciter coil's and/or internal to the CDI (rouge oscillators etc, who knows) and say the cap ends up with 400VDC, then you end up with 40,000VDC and so on and so forth.
At some point, with excessively high voltages, the insulation resistance rating of the rubbers/plastic etc all used to make the ig. components is exceeded, things start to ionize and arc'ing (Tesla coil style) starts happening all over the place...Super high voltages have magically ways of getting out/through stuff
I'd bet if you measure the primary voltage in the Ig. coils, its waaaaaaaaay high. Just a guess though
