If you quantize a track and a note is on the very last line of the pattern, this note won’t be quantized.
If delay < 80h, it should be moved a line (or more) to the top.
If delay >= 80, it should be either moved to the top of next pattern in sequence, to the top of the same pattern or removed.
I think you forgot this one. Just leaving the last note is IMHO a bit ridiculous… 
Maybe a brave decision required…
If it would quantize that note, it would process outside of the selected range. For single selections in the pattern editor (pattern fractions) this may be OK, but when for example quantizing a single pattern only, it should not change/modify other patterns (those patterns are may be used more than once in the sequencer too).
I’d keep the current behavior for now for the sake of stability - not changing something that we don’t have to change shortly before 3.1 goes final. As a compromise it could keep the last note in the selection with a delay of “FF” (nearly on the next line). But I’m not sure if that really makes anything better.
Hm, but IMHO it makes no sense for example if you quantize to 2 lines and then one note still is left on the last line, outside the raster. If the note delay is higher than 7F, I would prefer to have it deleted. I mean, who needs that remaining note, then?
EDIT:
Or, wouldn’t it be the most coherent, if
delay < 80h, move to raster
delay >= 80h, and pattern loop enabled, move to top of same pattern, NEXT to a note that maybe already exists
3.delay >= 80h, and pattern loop disabled, move to top of next pattern in sequence, NEXT to a note that maybe already exists
Err, the stuff above I wrote about delay value is of course rubbish, as u already realized…
But I hope u would agree that moving such note to the top of same/next pattern is much more logical than just leaving it untouched.
Guys, it’s ridiculous that the very last line in the pattern won’t be quantized… Can you improve? Thanks.