I’m gonna toot my own horn here, and slightly deviate to a related topic, “integrating spheres”. If you take a flashlight with a known/verified lumen count, you could put it through a series of diffusers. One to diffuse the beam, the second to scatter the diffused beam. Then, place all of that into a dark box painted with the darkest non reflective paint you can find, with a lux meter set up affixed to an adjustable platform. Tune the number readings to show what the lumen value is, and you will essentially have a pretty good integrating plane.
The logic is that if you take a cross section of luminous flux, you can accurately tell how many lumens came from a source.