"What is good enough".....
For me, +/- 3db is good enuf, but I have mine down to +/-2db except for a room dimensionally induced dip of 4/5db at 45hz (depending on room temp I think) which shows up in tests at 40/50hz levels, and this is referenced to flat FR at 200 and 1000hz.
Now for you, a couple of comments.....
Re test disc's. The test disc's are putting out 1/3 octave tones, the related frequencies, by number are not linear. For example, starting at 32db one octave higher is 64db, the next octave higher is 128db, the next octave is 256db, it just keeps on doubling, so by the time you get to 10K to 20K your deviding up 10K Hz difference into 1/3d octaves. The numbers make it sound big, but its not.
Rives is right about dips being caused by cross overs at or near the cross over points. They are benign, usually, because they are neither too deep or too wide to be exceptionally noticible. And just cause the speaker's cross over point is, say 200 hz doesn't mean its flat (or not) there or nearby.
Re high frequency roll off - this could be a result of test equipment as well as the distance from the speakers when the tests (yours and the manufacturers) were made. Usually we sit a lot further from the speakers than the manufacturer when the testing occurs and the high frequency roll off is going to be more prominent. That all of the instruments you mention sound fine is because all of their primary, and most of the overtones, are well below 10K. (O/T a great 'tone contol' is one that rolls the upper mid range (1500 to 3000Hz) and leaves the highs 'flat'.) What you might miss from the rolled highs is that, with a well set up system and speakers with good resolution, is detail in the recording of room ambience caused by the upper frequencies (overtones) emitted by the instruments.
FWIW, assessing the impact of irregular FR - One db differences are barely, if audible to most anyone except when you are listening very critically. 2db is barely audible under normal listening conditions. 3bd will make a very noticible difference. 6db can make it or break it. But, how the differences can change is even a 12db difference can make little, or no difference, IF it is very narrow and does not appear in a critical location of the FR spectrum. You might not really notice such a peak in the upper bass, lower mids where it would add a tad of warmth but not be clearly distinguishible, but at say at 2500 to 3000 hz, with a trumpet playing a solo it could put an edge on the sound that would take your ear off.
That was all just a stream of consciousness, so to speak, as a preliminary to this.......If you want some specific advise you might be well served by telling folks very specifically what your actual frequency spectrum is. When you talk about "75 and 85bd but with peak at 93db" at 184 hz what are you referencing the peak of 93db to? 80db at 1000hz, 200hz? What exactly?
For example, and just dealing with the bass, what I would say if I owned the questiojn, is that my room is 13x19x9, my speakers are 66 inches from the wall behind them, 9 ft apart, toed in until the axis crosses in front of my listening position 10 ft back. Referenced to 80db at 1000 and 200 hz, which are flat or measure:
80db 0db (flat)
200hz 0db (flat)
Seeng this, especially in graph form, helps to understand how big and where the deviations from flat may occur, and help determine whay they occur, and what you can do to fix them. Some times as simple as moving a chair or your speakers, or both in combination, less than a foot.
I hope that helps you a little bit. Something to think about anyway. Oh BTW, there are no quick and easy fixes, but by doing a lot of experimenting first with things as simeple as moving stuff about you can save a lot of money trying to find quick and easy 'fixes'. You just need some graph paper (a lot perhaps) and even more patience. :-)