Associate voices with rests

You can, and actually have to, start with the first rest of that measure (you can delete non-voice-1 rests later)
Select (any voice 1 note or rests anywhere in the measure), press N (starts note entry mode, switch to voice 2 (moves to 1st position in staff, if not there already), start entering notes and rests as needed.

(This site used MarkDown, so single asterisk or single underscore denotes italic, when doubled thes denotes bold)

There definitely seem to be some misunderstandings here, hopefully this will clear some of them up.

The buttons on the note input toolbar - note just the voice buttons, but pretty much all of them - duration, dot, accidental, voices, ties, etc - do different things depending on whether you are in note input mode or not. This is very much by design, in order to facilitate the very different things one actual does in these two modes. In note input mode, you are, well, inputting notes. In normal mode, you are not inputting notes, but making changes to existing ones.

So, pressing pretty much any of these buttons in normal mode will take what notes or rests are selected and apply the effects of that button to those notes or rests. Select five notes and pressing sharp adds sharp to them all. Select two rests and pressing the dot button adds dots to them both. Selecting a handful of notes and rests note and pressing a voice button sends them to that voice if musically possible/appropriate (again, this is not the usual way of manipulating voices, but is provided as a limited way of doing certain common edits a little more easily than the more general tools).

In note input mode, none of these buttons are normally expected to change anything whatsoever about notes already entered, because your whole reason for entering note input mode is to enter new notes. So these buttons instead set the stage for the next note you enter. Selecting a duration button doesn't change the last note you entered or any other note - it just means, the next note you enter will have that duration. Selecting an accidental button doesn't add an accidental to any existing note (although, inconsistently, it used to, but this was fixed a while back) - it means the next note you enter will have that duration. And the same for the voice button - it's never going to alter any note you already entered while in note input mode, but it will make sure the next note you enter is in that voice.

So absolutely, it's correct and desirable that these buttons are optimized to do the different things that are actually expected in these two different modes.

So, it appears you have an unusual case here where a voice has been entered only partially to begin with. Your example has a half note on beat three but no leading half rest before it. Presumably this happened as the result of having previously deleted that voice 2 rest. That's unfortunately, and I've long wished we nevever allowed this, but I know the Implode command will also do this in certain cases as well, and some people don't mind working this way. I always advise people, though, don't delete rests, make them invisible. The "holes" left behind are hard to deal with.

So indeed, the very special purpose for which the special move-individual-notes-or-rests-to-a-different-voice command was implemented not that long ago doesn't really apply here - it was never meant for this case, but rather for splitting individual notes out of chords and a few other very special cases that aren't easily handled other ways. So it has limitations indeed that come in part from how it was optimized to serve the specific needs it was trying to address, but in case, also from the unfortunate case of the deleted rest that left a hard-to-edit "hole" in voice 2.

I've tried to argue we shouldn't allow these holes, but enough people seem to prefer them to making rests invisible that I am afraid we are stuck with them. So instead, it's a painstaking matter of finding all the individual special cases where they cause problems and trying to find individual fixes for them all.

Anyhow, long story short, indeed, adding leading rests to incomplete voices can't be done in this particular way. If you end up in a case like this, there is a simple if non-obvious way to add back them missing rests: exchange voice 1-2, twice. The first exchange moves the incomplete voice 2 to voice 1 and fills in the gaps, because voice 1 is not allowed to be incomplete. The second exchange then takes this nicely completed voice 1 and moves it back to 2.

I guess more to to the point is, how did the measure get into this state, what's the desired end result, and wouldn't it have been much simpler to just enter the notes that way to begin with? That's been my point all along - the voice system really works amazingly smoothly if you don't try to do things indirectly in several steps but instead just enter the notes into the voices directly. It's possible to take the scenic route, but indeed, sometimes that scenic route will involve other details as well, because it wasn't never designed to be the direct route to the specific result that direct entry was.

As for notes being color-coded, I think most people would find it incredibly distracting to have their scores constantly displayed in color rather than black and white. But I could imagine an option to toggle this display on and off, feel free to make a suggestion in the issue tracker. Meanwhile, there is the colornotes plugin, not quite the same as it won't automatically update, but could still be useful at times.