That really doesn't look particularly sharper than the graphs I'm looking at with 0 smoothing
even the smooth lines on my graph, like the incline from 31hz-1khz are not entirely smooth, you can see jagged edges there because the pink noise file that’s used for measurements is, at the end of the day, randomized noise; it will create minor artifacts that don’t affect the result of the measurement but do make it extremely obvious if smoothing is being used.
Also, don't you mean 8k peak?
8khz peaks on IEC 711s are standardized coupler resonance relative to insertion depth, they don’t exist (unless the IEM itself has an 8khz peak in which case it would exacerbate it) in practice.
I don't see peaking at 6k
look at the peak after 4khz. this IEM is dampened as shit, so the higher peaks are suppressed by the inherent treble rolloff.