Deeper scale support for note input, visualization and RNA

I respectfully disagree with this assessment of my intent.

I do not want MuseScore to write the music for me, I just want it to help me and hopefully others in the difficult process of learning more music theory to feed their composition process.

From having tried the intuitive approach that you advocate (composing from listening and performing experience alone) in isolation for a number of years, I am convinced that there are serious benefits to complementing it (not replacing it, obviously) with the more analytical view that music theory training proposes. Here are some reasons why.

The intuitive approach is particularly prone to getting you stuck in a certain musical comfort zone. Patterns that you don't hear/play often do not get memorized well, so unless you take an enormous effort to avoid it, you inevitably end up mostly learning the clichés of the music you're most used to, and learning anything else very slowly and laboriously. You can try to avoid this through experimentation, but this is actually a very slow and thankless exercise since ultimately, there are not that many musical patterns that sound good, so I would certainly not recommend that to beginners. In contrast, formal musical theory training tends to go for the opposite extreme of enumerating lots of different musical patterns, all put on the same footing irrespective of how common or niche they are. This is often derided as wasting your time on lots of stuff you'll never use, but is also a great way to broaden your musical horizons.

Another benefit of musical theory training is that it gives you lots of names for musical patterns, that you can subsequently use to more easily discuss musical concepts with others who underwent similar training. If your only sources of learning are listening and performing, you will eventually learn a couple of these names, but very slowly and incompletely, so it's a lot harder to exchange with others and you're more isolated in your learning process.

And an important third benefit of formal training is that it gives you an analysis framework that you can use to more easily understand and learn from the music of others. It does so by letting you mentally break down other people's complex music into simpler components, study these components separately, and only then see how they combine with each other to form a cohesive whole. And it also helps by giving you mental shorthands for relatively complex sounding patterns (this is a scale pattern, this is a I-vi-IV-V chord progression, this is that particular flavor of counterpoint...). What all of this does is to make other people's work less impenetrable, by breaking down the wall of mental complexity into building blocks that are more digestible to your brain, so you learn more of it and more efficiently.

All this to say : I think there are significant benefits to learning to compose with a more analytical mindset. And if computers can help us at doing so by automating away some of the evil tedium of getting used to the music theory mindset (such as needing to constantly juggle with 12 slightly different versions of the same idea because transposition), I would argue that's a good thing.

I wonder if this is working as intended. If I switch to something else in that dropdown, nothing seems to happen, and if I click elsewhere and then click back on the key signature, my choice has been reset to "Unknown"...