Yes, the risetime when the triac fires is pretty quick, but I'll bet that still won't give the current levels you are measuring. Look at the risetime on a scope and calculate it yourself. I = C* dv/dt "I" in amps, C in Farads, dv/dt in volts per second

I still think your setup is showing you PEAK, not average current.

Peak current can easily go up with reduced duty cycle. Consider the fact that if you were firing the triac at 60 degrees into the cycle, the bulb would not be at full brightness. Since the filament would be colder than normal, its resistance would be lower than the normal operational value (tungsten resistance is inversely proportional to temperature). The peak voltage (at 90 degrees phase angle) would still be the same as before, and working into the lower resistance of the colder filament, the peak current would be a bit higher.

One other thing comes to mind: I'm not an expert in current transformers, but is it possible that they have a frequency-response component to their operation - that is, if the frequency goes up, their output does as well? If that is the case, the short risetime when the triac conducts would look like a high frequency signal and couple better than the normal 60 (50?) Hz stuff.

Whenever I need to measure current, I use a Hall-effect device. It has perfect isolation, near perfect linearity, doesn't cost much, and Doesn't have any "funnies" that I have to deal with.