Its a thought at least. The lux meters we currently use are not optimized for LED lights either for example, so it may be a simple matter of calibration.
If you are counting on the panel being ABLE to absorb all of the energy, assuming a known conversion efficiency, etc...it should be doable...even w/o the meter space.
For example, putting the light bezel down onto the panel and turning it on should give the same thing if the surface area of the panel hit by the light is able to absorb it all.
If the light emits too much for its "Bezel print" to absorb, then backing it up to make a wider circle of light on the panel until it CAN absorb it, yet loose no spill to the sides of the panel, etc, eventually, an optimum set up could be developed.
A solar panel uses visible light mostly, but the sun is sending IR for almost 1/2 of its energy, and UV for about the other half.
Considering that most LEDS produce little IR or UV, the beam from the LED might only have 7% or so of the sun's energy just based on wavelengths.
There's a gap in energy where the LED would need to hit the panel with enough juice to send current out the other end....and that combined with the wavelength issues makes me thing you might be able to actually just do a bezel down measurement with a conversion factor, etc.
I don't know the conversions between the light's energy and the panel's use of it though, but, it has some promise.
ISP = Integrating Solar Panel.