I guess I might as well throw in my 2 cents... should take us up to $1.14 or so.
Back before I blew out my HF hearing I spent a while building speakers and playing with crossovers - hobby, not professional. After a couple of years, my thinking was more or less :
1. For the things I could measure or easily hear, high quality components did not give noticeably different results from low quality components AS LONG AS the relevent specs were the same. I just measured resistance and capacitance; I didn't have the math to know exactly what to do with inductors other than measuring series resistance and staying with air core inductors 'cause I *did* understand them
2. Small differences in the values of the components *did* make a surprising difference in perceived and measured frequency response, more so than I expected. I don't remember the details, but I do remember that I was spending extra for more precise capacitors but was not spending for exotic resistors, which makes me think that somewhere around a few percent was the threshold.
3. The testing methodology I had was relatively limited - sometimes blind but not instantaneous A/B - so I had a tough time comparing things like soundstage width and seamlessness. I never felt that the speakers I made matched the best I had heard in that regard, although at the time I was probably underestimating the importance of the room.
So... bottom line for me is that if the crossover design is relatively complex or has sharp cutoffs, replacing the components with higher quality parts *could* noticeably improve the sound if the original components had sufficiently broad tolerances that not all speakers from the mfg sounded the same. Put differently, the benefit from the higher quality components would come from the tighter tolerances relative to the sensitivity of the crossover design even if the "higher quality" itself did not make a difference.
I don't understand enough about what affects the nuances of speaker sound, particularly soundstage and imaging, to be able to comment intelligently whether higher quality components could help there. Intuitively it seems like "higher component quality" alone should not be able to make a difference, other than tighter tolerances both new and as the components age, but I would also be interested in the results.
Oh yeah, I also learned that if I tweaked the crossover myself without either measuring FR or having other listeners involved I invariably "improved myself into a corner", where the sound was as good as I could make it but it sounded like crap compared to reference speakers.
Anyways, I think it's great that replacement crossovers are available for the M22s. The debate about the relative contribution of cabinet vs drivers vs crossover vs "optimizing across 'em all" was raging when I first got into speaker design 35-odd years ago and I certainly haven't found the definitive answer yet.
Finally, if anyone wants to visit Axiom for the test (assuming Ian has space in the listening room) but are on a budget, if you can get anywhere near Toronto by bus or whatever I would be pleased to provide low quality accomodation (albeit with good eats and good beer) plus transportation up to Axiom and back as required.