Do you have a source of this contradiction to Tooles research?
A "perfect sound" was never stated or the asked, the reference for any recording can only be what was heard in the studio and therefore
above is not true as for example a stereo recording is already mixed and mastered with and for the playback with stereo loudspeakers.
You mean your anecdotal claims of non blinded testing?
Erin criticised in his video review the 2-4 kHz dip and preferred it equalised.
Have done that and while some times I have some lust for such voicings (especially with some older/poorer redcordings) I usually keep coming back to flat in the lorn term.
With that said, a 3 KHz dip is probably the best compromise you can do in a stereo setup.
Linkwitz explained this in his blog
Electro-acoustic models
Acoustical and electrical models for the design of a dipole loudspeaker with numerical examples for the PHOENIX project.
www.linkwitzlab.com