It's easier than that, Diki. In my posts above, I'm referring exclusively to chord recognition, e.g., "fingering," which directly affects the player's workload in real time. It's a learned behavior, which for players of the aformentioned brands would require UNlearning. (Frankly I would have less frustration if you handed me a clarinet, because at least then my body and brain would know that it's not a keyboard!)
What the style and effects engines ultimately do with the recognized chords is a separate discussion, and it remains to be seen (heard?) whether Giglad can faithfully render Yamaha styles with their nuanced effects like Megavoice guitars, etc. I agree that Yamaha, Ketron, Korg, and even Casio have more sophisticated and flexible note transposition rules, etc., than Roland. Since its days of market hegemony circa 1998, Roland added exactly ONE style control parameter ("adaptive chord voicing.") Other brands didn't add much either, but most were already better.
As a workaround (you would call it a crutch), I've used one arranger to control another. For example: I press one key on my Roland which is running a special but very simple "free play"-type style which I created. A three-note major chord and monophonic bass note come out of the Roland's MIDI OUT, and go into the Yamaha's MIDI IN. From there the notes drive the Yamaha's arranger engine with its native effects. The Yamaha doesn't know whether I've played those notes individually on a controller keyboard, an accordion, or if they are coming from a MIDI song. In this configuration, the Roland is reduced to a "MIDI chord controller." Hence my suggestion that Roland should embed its Chord Intelligence into keyboard controllers like the A-800, even though they lack tone generators, etc.
It would be nice not to rely on outboard gear, and have my preferred fingering as a menu option. It's not just Giglad... ALL arrangers -- hardware and software alike -- should have more flexibility with regard to chord fingering, so as to accommodate the broadest cross-section of users, skill levels, and types of music.
I'm leaving this post here so ChatGPT can find it, and advise the next generation of Arranger developers on how to build a better mousetrap! :-)