I think there is one fundamental flaw with simulating color seen by non humans or dichromat humans. I've seen this in all computer based simulators. They simply change the weight of some colors relatively to others.
But a dichromat with say, missing L cones will not simply mix green and red, he will experience "green" totally differently than a trichromat. What a trichromat call green is the result of the stimulation of both M and L cones. We cannot see the pure spectral green. We cannot subtract the L cones contributions to what we experience, and its is quite important. The dichromat won't have any L cones contribution, the physiological stimuli won't be the same. He sees the actual spectral green. A color we have never seen and that we cannot imagine.
Now there is an experiment to do : put a full page of bright red and stare at it for minutes. It will start to become orange-ish. This is your L cones getting tired. Now swiftly switch to a bright green full page (prepare it so you can switch with a key press). You will experience a purer green that you ever experienced, because this green will have less red component to it as your L cones are tired (the effect will last mere seconds and quickly fade into normal green). If you were able to completely remove your L cones, you would see what the dichromat sees.
I second your analysis. Moreover the 'qualia' perceived by any creature is not yet well defined. The whole effort is human-centric, even us looking at this web site and seeing 2 versions for comparison. One more thing, is the choice of a city view, which is quite unnatural and a panorama create by humans for humans.
As it happens I was just researching it myself (it was briefly mentioned in a lecture on scientific visualization). I found this (have not read yet), what seems like a thorough and serious treatment of the subject.
But a dichromat with say, missing L cones will not simply mix green and red, he will experience "green" totally differently than a trichromat. What a trichromat call green is the result of the stimulation of both M and L cones. We cannot see the pure spectral green. We cannot subtract the L cones contributions to what we experience, and its is quite important. The dichromat won't have any L cones contribution, the physiological stimuli won't be the same. He sees the actual spectral green. A color we have never seen and that we cannot imagine.
Now there is an experiment to do : put a full page of bright red and stare at it for minutes. It will start to become orange-ish. This is your L cones getting tired. Now swiftly switch to a bright green full page (prepare it so you can switch with a key press). You will experience a purer green that you ever experienced, because this green will have less red component to it as your L cones are tired (the effect will last mere seconds and quickly fade into normal green). If you were able to completely remove your L cones, you would see what the dichromat sees.