Does Oratory use an identical measurement rig to Harman's? If not, and they're using straight Harman-target values on their charts, you're not actually EQ'ing to the Harman curve. Rather, you're EQ'ing to a corrupted Harman curve that is altered by the transfer function between the Oratory rig and the Harman rig.
So, it's hard to say the Harman curve is faulty without using headphone measurements used on Harman's rig.
While this is true I'm not certain that the differences are that huge below a few kHz for large, open over-ears, as only the pinna is different. In Harman's study on leakage, limited to the spectrum below 1kHz, the older, stiffer GRAS pinna and the custom pinna didn't affect much the HD800's response :
Headphone leakage effects can have a profound effect on low frequency performance of headphones. A large survey, including over 2000 individual headphone measurements, was undertaken in order to compare leakage effects on test subjects and leakage effects of the same headphones measured on a...
www.aes.org
I would expect the most significant differences to occur above a few kHz.
But IMO the combination of the headphones transfer function and the sample variation issues make all of this a bit of a moot point. As an illustration of that, the sort of deviation that I see with my own samples when equalised to Harman's target per Oratory's profiles in the ear canal gain region when measured in situ, even with different microphone types, is
higher than the difference between the various revisions to Harman's over-ears target anyway. Or, at lower frequencies, the variation is at least equivalent to the variation in preferences expressed in various studies or to the differences that were noted in the 5128 vs. GRAS + custom pinna study, particularly when including headphones reliant on seal to deliver the designed bass response.
At lower frequencies this is visible in the same study above. The K550's on-head response on the eight human subjects has little to do with the ear simulators, even when using the custom pinna.
BTW this is why I don't know whether or not I like Harman's target above 800Hz, because I can't know what it sounds like given these variations
. However, ANC headphones with a solid feedback mechanism, no volume dependent EQ, and tight tolerances (such as the AirPods Max) will deliver an exact dB value at your eardrum for a particular digital value at lower frequencies, so provided they were properly measured (which I don't think is a significant issue with sweeps in the 50-800hz band when ANC is turned on for the APM), it's one way to know for sure what the bass shelf sounds like.
So as long as Harman's target is used as a reference point, a guideline, or as part of a statistical exercise, I don't see it as much of a problem to apply it to GRAS rigs without the custom pinna as we're not dealing with levels of precision that would make it matter IMO. And if it's used as a way to dumbly calculate a score to evaluate a pair of headphones in particular, well I don't consider it a particularly scientific endeavour anyway regardless of the pinna used given the HPTF issue, so...
Oratory designs his profiles' filters at lower frequencies to enable fairly easy fine-tuning of the response, particularly for open over-ears like the HD800 with a fairly smooth response in this part of the spectrum, but some of his profiles are less flexible at higher frequencies.