I thought about a lipo testing scheme for some time. If not high constant current load; how about measuring transient response to varying loads? I'm almost certain that if we use impulsive loads and moderate constant current loads in a repetitive pattern it will work. Then graph or create formulas to translate the results to usable data for grading cells. You can also change the load patterns to grade cells for different classes of racing. I can probably build this machine and write the software for it but not out of my pocket! hint hint Mr. Black!
I've thought about something crude in this fashion using some type of pulsing relay (or something like a Heavy Duty Automotive Flasher unit), hook the battery up to something that would pull a constant 15 amps discharge, and have the pulsing relay kick in an additional 25-30 amps for 1-2 seconds at a time, and do this over a timed run period too.