For speakers a flat anechoic response, with an even dispersion is a well established target.
With headphones you would want to match the eardrum frequency response of the speaker above, in a real room. The outer ear has complex acoustic interactions vs. an omnidirectional microphone used for speaker measurements. That poses a challenge, on top of that headphones on their own can't reproduce inter-channel effects. It all has to be approximated with frequency response by proxy.
So Harman went ahead a did a good job at quantifying the required FR. Fantastic. But before even considering everything that diverges from Harman to be deficient,
consider that significant mismatches between eardrum responce and ear simulator do exist. significant unit variation exist. Few listeners strictly adhere to 83dB SPL, sitting down in a quiet room. Definitely consider all of that, show that you understand the real world variables, before even suggesting someone is backwards for not being 100% onboard with Harman in every single situation. Unless you want to give sceptics a valid objection. I sure don't.