It's definitely not just the magnetic switch! I recall Petzl zooms throwing compasses off many yers ago, and I demonstrated the effect on a compass from a Petzl Tikka on Saturday, neither of these have a magnetic switch. Compasses are very sensitive instruments, and it does not take much of a magnetic field (from electrical equipment, permanent magnets or even ferrous metals) to alter the local field enough to noticebly affect the reading.
For this reason I always remove my helmet when taking readings, and illuminate the dials with a small low powered light I take in specially for the job. This is held far enough away (about a foot is enough) that it won't affect the compass (remember that the field strength is related to eith square or cube of distance). This should elliminate your illumination screwing your survey up. I do like Sam T's method of using small glo sticks - one I must get round to trying.
Back sights will tell you if you have made an error, and as long as any magnetic deviation is constant, the included angle method can be used to correct for it. However deviation caused by a head mounted light is unlikely to be constant for two readings, even a slight difference in the relative position of compass and light will alter the deviation. This is unlike magnetic anomalies due to the rock the cave is in, where the anomaly is constant at a given point (i.e. survey station).
Call me a sceptic, but I would like to see some hard eveidence before shelling out over £200 on a new light on the basis that it won't affcet my compass!