If I understand the principle, you can delay a pulse but not advance it. Don't think retarding the spark would be worth anything. Theoretically the CDI induces delay since the capacitor needs a certain amount of time to charge. Charge time ought to be a factor of capacitance, more mfd = longer charge time. Remember we are talking milliseconds here, so it's a totally rotten place to mess around.
I apologize for not looking first, but how is the inductor magnet fixed to the rotor? Is that the keyway you mentioned, or is the key on the rotor itself? Is there enough space to add a rare earth magnet ahead of the existing one? You'd wind up with a longer pulse and (maybe) a longer discharge interval from the CDI. Dunno if that would mean a weaker spark over a longer interval or if it would make a hotter spark due to the extra charge time on the cap. Hmmm....
Doc