It is very hard to imagine how a conventional magneto could output short bursts of around 122 MHz to activate the ARCAL receiver, especially since it is presumably in normal operating condition. I would be inclined to write this off as a troll or joke, but Magman's posts have been straightforward and helpful in the past...
I would suspect a problem with the ARCAL receiver. It is supposed to detect a discrete, highly accurate carrier frequency, in bursts of 3, 5 or 7 within five seconds (depending on J or K type), so it should not be easily fooled by ignition noise of any sort. It would be interesting to know if any other airplanes are affecting this field. And apparently, it only happens when the airplane is in flight, throwing another factor into the problem.
The airport runway lighting system is, if it's like the one I used to maintain, a long string of transformers, one buried under each light abut 20" down, all wired in series in a 600-volt circuit. Each transformer drops that voltage to 4.5 volts for the bulb, a low voltage to try to avoid big sparks that might set fire to spilled fuel in an airplane splatters itself on the runway and breaks a bulb.
That big loop of wire all the way around the runway and taxiway and ramp could act as a huge antenna, picking up RF and feeding it into the electrical supply near the ACAL receiver/decoder box. But, again, that receiver and its decoder would have to be pooched somehow.
Electronics can be tricky. We ran into a problem with the alternator going offline in flight, right after we took off. It turned out to be a leaky Com antenna connection in the ceiling of the airplane, where moisture had corroded the bayonet nut's grounding. When we made the (uncontrolled field) call for takeoff, Com RF radiated out and was picked up by the airplane's wiring, creating sharp little voltage spikes that the ACU (electronic alternator regulator and overvolt sensor assembly) would interpret as overvoltage, and the ACU would shut the regulator and therefore the alternator off. Those electronic doodads work on tiny currents compared to the old electromechanical regulators and 50-year-old overvolt sensors, which needed much more current so that escaping RF wasn't enough to bother them.