I recognise the snip, as I use deCoda all the time, and for some tasks it’s far more useful than Scaler. However, the reason for that is, IMHO, the applications are as different as chalk and cheese, and (again from my very personal perspective) i see little overlap between them. I can see where you are coming from (and they are valid points) but it’s hard to see how either would move towards the architecture of the other.
{1} The first key significant difference is that deCoda is wholly focussed on the audio domain (‘midi’ is mentioned only twice in the manual, neither with a functional role) whereas Scaler is wholly focussed on the midi domain. They deal with different things. Scaler works with audio only to enable audio to midi transformation, and hence it is unlikely to offer any of the things deCoda does. The reverse is true - deCoda can’t do everything (actually, more or less nothing) that Scaler does.
{2} The second major difference (and the reason I am frequently using deCoda} is that time is not recognised by Scaler (I suspect with good reason), only changes in input. However, detecting timing drives much use of deCoda, as people use it (for example) to get song sheets of tracks want to play. The only real midi aspect of deCoda is that it can produce and incredibly badly formatted print of the same.
{3} One big difference is that deCoda dumbs down the chord recognition. By and large, it senses only triads, and occasionally something like a 7th. Scaler is significantly better here, sometimes over the top for some extensions.
I don’t see how your suggestion of editing a composite track waveform in the audio domain would work. The main window shows composite audio of potentially n different instruments overlaid with multiple processing effects. Imagine some section has been detected as G major within an audio stream consolidating three guitars, synth, keyboard , sax , drums etc, all playing several parts. How could you take out that G major stream and replace it with an E minor one? Once you have mixed paint, you can’t un-mix it.
{4} Finally, there is ‘philosophy’. deCoda is not (in my view) a composition tool, and is wholly reactive. Conversely, composition is the heart and soul of Scaler, its very raison d’etre.
Don’t get me wrong, and please don’t take this as a criticism of your post; its not intended to be. Audio detection in Scaler needs work, and the devs know that. I see where you are coming from, but I can’t see how it can be credibly achieved in the way suggested. However, there is a reason why I turn to deCoda frequently rather than use Scaler, and that makes your point very valid.