Why don't people use a light meter, as silverFox mentions? I hear so often that an integrating sphere should be used. But for a small shop or hobbyist, I cannot see why there would be a functional difference in the result -- just more labor in taking the measurements.
It would work like this: take intensity measurements at regular angular intervals in a half circle, in the plane of the light. Go from behind the light around the side of the bezel to the front. Then numerically "integrate" these values over strips that make up an imaginary sphere, multiplying each intensity by the appropriate area and summing the total. Convert units, and poof, you have output.
The catch is that you have to do this where there is not a reflected light problem, i.e., don't do it in a white room.
I can see why you would pay $10k for equipment if you are Surefire, but the above approximation should be perfectly adequate for informal purposes. You shold be able to get lumens output that way.
Scott