fix #92331: following chord deleted after voice change

[MarcSabatella]

See my comments in https://musescore.org/en/node/92331#comment-416681. I might be confused about the actual purpose of expandVoice() and the various makeGap functions, so I am trying to be explicit here in my comments. There the cause of the bug is that expandVoice() is supposed to fill a gap at the current segment as well as preceding the current segment, and that makeGapVoice() is correct to assume there isn't already a gap present. expandVoice() was sometimes filling a gap at the current segment, sometimes not - depending on whethere there was also a gap preceding the segment or not. I have taken out the two lines that prevented expandVoice() from filling a gap at the current segment in the case where there was no gap preceding this segment, and this fixes the reported bug. So far, I don't see any ill effects.

I have also added comments to make the function of expandVoice() more explicit. Again, these are based on my assumptions about the purpose of the function.

[lasconic]

It looks correct to me.

I guess it would be good to write some tests for expandVoice() or the callers... any idea?

[MarcSabatella]

If my understanding of the function is correct, I could probably write an mtest just for expandVoice(), to see if it does what I expect in a variety of cases. Maybe a separate one for makeGap(). Between the two of them, they are called a lot - well, maybe only a few direct call sites, but a lot of different paths to reach them. So we are already doing some tests by testing those commands.