A lot of things are being confused here. Speaker measurements, Harman curve for speakers, Harman curve for headphones, and so on.
The Harman curve is fundamentally based on listener preferences. I don't know why people are so resistant to this idea, as it is called out clearly in the research. The curve is defined by what people preferred. It seems many people are not basing their opinions here regarding the actual research, but rather what they wish the research said. How very unscientific of you.
As to speakers measured in an anechoic chamber, it seems to me you couldn't pick a worse baseline measurement for audio reproduction, as it doesn't meet any criteria. An artist could not possible intent their listeners to listen in an anechoic chamber, a sound engineer does not mix in an anechoic chamber, a listener does not listen in an anechoic chamber. How many listeners even know how music sounds inside an anechoic chamber? I'd guess under 1% of listeners. Perhaps even 1% of artists themselves. So if we are trying to say that accuracy implies an artist's intention, an anechoic chamber is irrelevant.
As such, there is a very real basis for attacking the fundamental definition of neutral in the Harman studies.
That said, even if we assume such a baseline, people's taste differed from a totally flat response in Harman's research. This is also explicitly called out in the research. People didn't like acoustics that generated a flat response, even when room corrected for a totally flat response. This is the basis for the Harman curve - people preferred something different.
Now here is where the sleight of hand comes in, and marketing takes over. Harman wants to redefine this Harman curve as what natural means, because they want to say, that the curve fixes inherent problems in the speakers and the room to give a totally "natural" presentation. You can see how this is a rather loaded assumption, because they already knew how to correct the room for a totally flat response - and that isn't what people liked.
I encourage people to read the following as well (pay particular attention to the figure on page 17, which shows Harman curve deviating from a room corrected flat response):
http://www.aes.org/tmpFiles/elib/20170625/17839.pdf
Finally, Dr. Olive's research showed that in the context of headphones, people preferred yet another Harman curve which deviated itself from the speaker Harman curve which, to remind you, deviated itself from how a totally flat frequency response was defined (again, if you are paying attention, we are using the dubious assumption that the anechoic chamber is the ultimate ideal for flat and neutral).
In short, by making the jump from "the Harman curve encapsulated listener preferences" to literally "the Harman curve is the objective standard for natural sound for all people everywhere" is a rather large leap that would please Harman's marketing department very much.
Point being, if you truly want to be objective and scientific, limit yourself to what the actual science is saying.
Any statement that presupposes the Harman curve as a target is a subjective value determination.
"I subjectivity choose the Harman target for headphones as my accuracy target, because I subjectively believe that preference consensus to the target threshold of 87% of listeners is enough to define said accuracy target".
It's subjectivity all the way down folks.