All chord recognition divides its recognition into specific windows. Some to the 1/8th, some to the 1/16th, etc.. But to my knowledge, there's no quantization done to it. You either get it right (and most of us do with no input quantization) or, if you get it wrong the first time, every CS subsequent iteration of the loop had the same exact 'glitch'.

With the problem of syncopated chord changes and variable swing values, I'm not sure exactly how a CS could ever recognize a glitch if it saw one...

The quantization is useful if you are using the CS to generate SMF Parts to be further edited in a sequencer, because you could stop and edit those at your convenience. But during a performance, It's going to be hard to tell a CS in advance what quant value and swing factor you are going to need.

From an early look at the BK-9 manual, it seems some post editing to the CS might be possible, to generate SMF sections for song construction (not 100% sure yet, though) without the tiny glitches in, and if stephen.hazel can manage a simple 'record CS' and 'Load CS' in SMF form, some easy editing might be possible.

But live input quantization I am sure could be left off until the project is working, and more advanced features thought of...