I was looking for and FR on those, but could not find one. That is exactly how that series of speakers sounded to me. However it is a pretty good FR.
The OP is talking about the series 1 which were sealed and not ported.
I also owned a pair for a while, and they were a very good speakers in their day.
The slight rise in energy is associated with cone breakup as the speakers start to beam in that area.
But the OP almost certainly has come across a better speaker then he had had before.
I think it's from the tweeter, not the mid, because the off-axis plots show beaming down at 2kHz, and the waterfall is very clean through that crossover region. The tweeter's lobe is also very broad in that 3-6kHz region, compounding the issue a bit, depending on room dimensions and acoustics. Fwiw, my 801FS's shared this trait too.
Back in the early 2000's I was able to listen to N800's and Matrix 801's side-by-side and the N800's were definitely the smoother sounding of the two, even with a higher measured treble energy above 8kHz. and the N800's don't sound substantially different from the 800D / 800D2 imho.
I agree though re woofer-mid integration issue, although it can be ameliorated with positioning and boundary reinforcement, but the problem doesn't seem to exist in the latest D3 series, although there can still a slight tonal imbalance in the upper mids due to the combination of high crossover and the +180 degree time alignment of the tweeter which further reduces off-axis energy in the 2-3kHz region.