Crossover point of 5.25” vs 6.5” in two way bookshelf

If all else is equal, the 5" would tend to have better HF response, but it’s not an absolute.  It could be a typo or simply a variable due to other factors with the drivers.  Either way, to allow the sound waves to propagate as a single point source, a 3100hz crossover would ideally need the driver centers to be around 4.5" apart or less, which is a bit of a challenge.  A 2100hz crossover is a bit more generous allowing up to ~ 6.75" between the driver centers ideally.  There are always tradeoffs to consider between coherency, dispersion, frequency range, distortion, comb filtering, etc.  Hopefully they've done their homework and chose the best situation for both.  

All these 2 ways are a compromise for the following reason...

A small driver covering around ~1000 to 1300 hz and above is an important concept, as there is a lot a spatial information, resolution, detail, harmonics, etc that can be preserved by doing so.

Small drivers playing that low can only come from a compression driver/horn speaker or tekton style of small driver array. Some 3-ways also do it in an intelligent manner. For, example, a small ~3inch very high quality zylon mid driver takes over at approximately 750hz and above in a yamaha ns5000, etc. The PS Audio FR10 is another example, a low mass ribbon midrange taking over from 550hz onwards...

This can all be demonstrated in a a/b with what to listen for in tracks. (not just the science of small driver impulse, low mass, etc)

In general, it is best to avoid these 2 ways. 

I have some eminence 6.5 inch alpha beta speakers go to 5 k without breakup.depends on how there designed. Shorting rings neo magnets ,underhangs, x max , vas, voice coil windings, compression,wattage ect. I would bet sonus faber has a fantastic sound room like jbl to do testing.i went to Phoenix AZ  lmc and listened to one of thier top models on burmester and big mac stuff. It was under pink floyd Pic of the prism or dark side of the moon. Well over 100k for that. Too bad I dint play for the suns I could have bought it. Nice people at lmc in the big city great stuff and one of the largest mcintosh  burmester dealers around. Enjoy the hunt. The ribbon planar have less mass respond faster. Ps audio had there ribbons made of teonix just under beryllium for modulus. Yamaha new material like teonix textreme is light quick response and push cone breakup out further.i still have the beryllium  ns 1000 yamaha as it's a great design. Final acoustics now has thier whole panel out of textreme. Jbl makes some of thier horns out of it 3800 model. I have a few of the horns emenince made of textreme then b& c bought them they quit that line. Enjoy the hunt and the science

I am not sure I agree with OP that the specs offered by SF seem backwards. A larger diameter woofer should beam at lower frequencies but also allow for higher power inputs. The tweeter should Ideally cross lower to preserve wide dispersion before breakup or beaming in the woofer occurs, but that stresses its power handling.  If SF is using the same tweeter then the higher crossover frequency is probably intended to reduce strain on it at the higher power levels expected from the larger woofer and enclosure.  What this tells me is the 5.25” model is probably a better option if you don’t need more volume or bandwidth than it can comfortably accommodate. 

Just two things.  One, there’s lots of excellent 2 way speakers with 6.5" class mid-woofers that sound really good.  Second, remember that the tradeoff for a low cross point is also reduced dynamic range and higher distortion.  It is a "compromise" no matter how you do it, but so is every single speaker out there.