An SCR datasheet shows trigger voltage changes from 1.0v to say 0.5v from 25C to 100C The thermistor is a shunt to ground here, it's hard to see because ground is not easy to see in schematic, it's the green coil wire. So when NTC heats up it's resistance goes down and it pulls the signal down to match. We have 3 signals combined to trigger SCR. This F15 to start, then Advance Pulse wave to advance before F15 with higher RPM. F43 acts as a limiter to avoid advancing any further. It's a complex analog circuit, so it's somewhat hard to speculate on it without testing, but I suspect the Advance Pulse overtaking the F15 signal is the most sensitive part that might benefit from the temp compensation.