I don't think we should exaggerate the problem with that dip at 2 kHz. Will anyone even notice that small dip when listening to the overall tone of a speaker, it's such a narrow band that is affected and will therefore hardly be noticed while listening to music that mostly contains instruments with much broader frequency response? It's not like listening to a frequency sweep where every single minor dip is easily heard. I don't think such small deviations are something you would be listening to while concentrating on the overall balance of a speaker.
1. Toole and his team came to the conclusion that most listeners prefer a speaker with a linear response.
2. If number 1 is true, we can assume that most mixing engineers also prefer that linear response, and have probably set up their sound systems with that in mind (except for the bass response which will probably deviate the most from the norm).
3. When the mixing engineer is making adjustments to the music mix, he will most likely, at least for modern productions, use EQ for every single instrument until they sound exactly like he wants them to sound in the mix. Most stereo flaws will therefore most likely and indirectly be addressed even if the engineer doesn't have any deeper knowledge of those particular stereo flaws, he just mixes the music until it sounds like he wants it to sound, and calls it a day.
4. That leads us to the full circle. If we end consumers have sound systems with a linear response, most recordings should sound fairly well balanced to most people. But there's nothing wrong if some of us have other preferences that don't go hand in hand in with "the middle of the road" as it seems to be with the OP of this thread. I prefer more bass than Harman's "trained Listeners" target curve, and opposite to the OP, I prefer a slightly forward-sounding midrange instead of a laid-back sound. To me, most music sounds more engaging and enveloping that way.