100 kHz was off the top of my head, as well. That's not critical by any means, but you'd want to avoid frequencies that could interfere with radio communications. LIke submultiples of 455 kHz, for example.
One factor would be to choose a frequency that would have at least several full cycles during the spark 'ON' time. For example, an engine at 6000rpm turns at 100 revolutions per second. If the spark was designed to operate for just over half a revolution, that's 1/200th of a second (5 milliseconds). I guess at 10kHz that would be 50 cycles, well over what I'd think would be a minimum, so that would work. The transformers are pretty efficient over a wide frequency range, so you'd have a lot of latitude.
I would definitely use coaxial cables for the spark plug wires in such a system, to prevent RFI and to transfer energy as efficiently as possible.
As far as 'spark resonant frequency' goes, I can't imagine the system would have a high enough Q to make resonance effects significant. The spark plasma itself is a fairly lossy high power resistor, and that would load the circuit enough so that resonance would be pretty broad.
Unless you have something else in mind by that term?
The Pulstar plugs evidently cause a faster growth of the flame kernel due to the intensity and speed (risetime?) of the spark. That's what the pictures at the manufacturer's website seem to show.
Supposedly the faster flame front means more complete combustion. If your engine is already burning the mix thoroughly due to good turbulence, homogenous mixture, and so on, this plug might not make a noticeable difference. I bet my old iron head Harley would pick up some benefits...
