The problem is: you have taken the studies and tried to use them as proof that this particular speaker is not accurate, you cannot conclude that using those studies. You can't use the study to say "oh, this speaker isn't high fidelity, just look at the curves". You can say that under blind listening conditions, some experienced listeners may prefer some other speaker that has a flatter FR response, but that doesn't mean the CM1 is junk or would fail completely under a blind test, and is automatically not "high fidelity". The study makes some general design principles that correlate sound qualty and objective measurements in an anechoic evironment. It's like saying "if you orient your kitchen this way, we found that cooks were more effective in the kitchen on average", but that doesn't mean there are no exceptions to the rule. You can have a speaker with some notable FR anomalies, and it could still sound fantastic because there are other technical aspects that it does very right. Just as John Tchilinguirian said, there is increased understanding of correlation between measurements and sound quality, but there is still more ways to go.
What is a high fidelity speaker anyway? Is a speaker that has a very flat FR curves but worse phase linearity better than a speaker that has great phase linearity and poor FR response. How do you determine which speaker is "high fidelity"? I know Thiel would also tell you that is a very important perforamnce characteristic and they design their speakers to have good phase linearity. What about cabinet resonance? Maybe instead of outputting a true 50-70Hz signal, some of the flatness in a FR plot is due to cabinet resonance putting it's own sound in the output? How does that fit into what is considered an otherwise good FR plot? What if a speaker that has good FR response have a slightly poorer shape in the impulse or step response? Is that still a high fidelity speaker? High fidelity is just a term that equates how real a speaker sounds, but realness is not something you can determine by noting that a speaker has a 3dB more dip at a certain frequency that another speaker. It's by listening, and running the speaker in a real world environment with real material. Because most music is EQ'ed, processed anyway, on studio monitors that have their own FR anomalies, and your own room has FR anamolies that easily swamp the minor deviation of FR shapes, you cannot predict whether a speaker is more real there another in the final reproduction of music by looking at a two dimensional measurement. If you think you are going to plop your speaker with better FR into your own room and it will automatically sound better than my CM1, I've got some cheap property to sell you
I can tell you that the CM can sound fantastic, and that's even put against the sound of real live instruments (within the limits of a very small speaker). To me, it passes some of the personal stringent tests that I use to determine sound quality. For sure there are some things it doesn't get right, but like many things, it's a tradeoff. To me, it gets some things right, especially in the timbre of the treble and lower midrange that some other speakers don't get quite right. There are other people that might disagree, like cschang or jonomega, who have heard the speaker, and who seemed to have listened to many speakers, but I can respect that opinion, as an informed opinion backed up with experience.
There's always people that come into this forum, and spew out a bunch of charts to prove their points, but it's understanding how these studies relate and correlate to the final sound is what's important. I've read most of the Harman white papers, and I've even visited the NRC facility before, so I'm not stranger to the science. Science is a tool, and I use it to help me save time and rule out things that it would be hard to determine manualy, but ultimately, the sound quality has to fool my brain into thinking that the singer is right in front of me.