Ok, let's revise this just a little so you can use a transistor for your discharge.
A 2n3055 should be ok if you don't let the cap voltage go too high. Wire the transistor collector to capacitor positive and the emitter goes to battery positive.
Now, you need a good trigger pulse for that transistor (sudden turn-on).
Make a small winding on something like a sheetrock screw. About 200 turns of #26 or something around that size.
You can turn a screw down into a small block of wood until only about 1/2" is left standing above the surface. Then wind a bunch of wire on there until it looks like a little "hocky puck".
This is going to be mounted so that the wheel magnets pass by and generate a little sine-wave pulse.
Connect a battery across this winding to find the correct polarity. You want it to PUSH the rotor magnets away when you stick the battery across it. Mark the pos and neg leads.
Connect the negative coil lead to a 25ohm potentiometer. Then connect the other potentiometer lead to the transistor emitter (pulse sensitivity adjustment).
USE AN SCR (not a triac) between the trigger coil positive and the transistor base.
The scr with conduct in reverse, allowing the negative shut-off pulse to pass to the transistor.
Connect a 3v zener diode anode to the SCR GATE and the cathode to positive trigger coil lead.
So, what's gonna happen?
The trigger coil is pulsed by the rotor magnets. The sinewave pulse goes postive and then negative, very quickly. Then there's a gap until the next magnet comes in.
The SCR and Zener keep any current from hitting the transistor base-emitter junction until it's well into the HARD turn-on voltage area. This will dump the capacitor and then the negative going side of the sinewave shuts the transitor off.
You'll probably want to heat-sink the transistor.
Start with max resistance and lower it to where it produces the best voltage push on the secondary battery.
No more sparks.
You can connect more transistors DIRECTLY in parallel. I've done this with no emitter resistors to "match" the transistors since the "on" pulse from the coil is already pretty high when it hits the transistors.
Probably more that you could do to improve this circuit. But, I'll leave that to you.
Enjoy!