I am not super hardcore about it. My stance is that you must listen to any correction above transition and if it sounds better, by all means keep it. Harman's own research with B&W 802 speaker by the way found that correction above transition was effective.
It compensated for a dip in its response that is due to poor directivity -- exactly what is NOT supposed to be correctable by EQ -- but subjectively it was. See AES paper, The Subjective and Objective Evaluation of Room Correction Products:
"As a final note, there has been much controversy
related to whether room corrections should attempt to
equalize the loudspeaker/room above the room
transition frequency (about 300 Hz). These
experimental results provide evidence that broadband
loudspeaker/room equalization above the room
transition frequency can yield positive benefits, even
when the directivity of the loudspeaker is not smooth or
constant. In these tests, both preference and spectral
balance ratings improved by filling in the 3 kHz dip in
the loudspeaker’s sound power response. The uniformly
distributed broadband acoustical treatment of our
listening room may be an enabling factor in achieving
good results with broadband equalization by minimizing
the room’s spectral tampering of the reflected sound
produced by the loudspeaker. More perceptual research
in needed to better quantify how the performance of the
acoustical treatment of the room affects the sound
quality of broadband room correction."