The Harman target is aiming for subjective preference. If you're looking for acoustic and measurements, this is not it. If you're looking for an averaged of preferred target, this might be your best bet. While real acoustic and personal preference will logically and hopefully find some common grounds, it's important not to confuse them and rely on a subjective curve to explain objective measurements.
here is a popular graph(not just because I spam it everywhere
), giving some idea of the frequency response impact that different parts of the body have on incoming sounds:
From this paper:
https://www.audiology.org/sites/default/files/journal/JAAA_08_06_06.pdf
Of course different people have different body shapes and sizes, so this should really serve as a rough estimate of the mechanisms at play. With a sound source coming from a different direction you'd get a different graph(see HRTF).
Someone with a different ear canal wouldn't get a first resonance at 2.7kHz like shown in the graph. Let's say I have a longer than average ear canal, then using the oversimplified closed tube model like they do at first in the paper, we get that the resonance frequency is related to 1 over the length of the tube(ear canal). So even without bothering with calculus and the speed of sound, we immediately get that a longer ear canal will have a first resonance at a lower frequency(first because we also get a boost at odd harmonics, 3X, 5X the first frequency for an actual tube, it might be higher for a real ear and that could be why some couplers show boost where you don't hear one at all).
And because of the shape and size of my ear canal, the amplitude of the boost might also be different.
And so on and so forth, you get the idea. Different people will be used to hearing external sounds more or less in their very own ways with their own custom frequency response. they still for the most part have 1 head 2 ears, etc. So I keep insisting about custom, but obviously that customization still follows main directions common to most humans. Sorry if I seem to constantly contradicting myself because of that, I don't know another way to present it.
Now to try answering some of your questions about what to aim for as "good" IEM FR. When you put an IEM into the ear canal, you send sound from a source that bypasses a bunch of those natural filters applied by our body. For audio to feel somehow more natural, we'll probably wish for a signature at the eardrum that comes close to what we're used to hearing from real people and instruments at some distance around us. And to oversimplify again, perhaps to a fault, this likely is why you'll want IEMs that have those changes baked in their signature. It's also why we don't want or expect a flat RAW response into an ear simulator(not even in an actual simple tube+mic, because it will still at the very least act like a close tube and resonate at some freqs). The measurement rig you use will also impact the final FR in that way and others, making it perhaps harder to even know what we're getting on the graph of a cheap rig like those I have.
In short, it's easy to make generously vague models for the right FR, but it's a real PITA the moment you try to become more accurate. I hope that with this I've shared some of my famous optimism.
If the previous paper and the many others on those subjects are not your thing(yet), you can join in here
https://www.head-fi.org/threads/ety...phone-for-your-ears-and-your-couplers.908512/ and see how we often try to reinvent the IEM-measurement-wheel between curious but not so well informed amateurs. The tour for the er2se is still on if you're interested in checking your rig with what the others got when measuring the same IEM.