decide how to read two same-named voices on a staff that meet without a backup in mx::impl #506

Written on behalf of the user by AI...

#492 (for #425) taught the reader that a change of <voice> text between two notes, with no <backup>, starts another voice. Within one staff, between two <backup>s, notes are split into voices by their <voice> text.

One case is left open: two voices with the same <voice> text on one staff that meet without a <backup> are read as one voice. The text is the only thing the reader uses to tell voices apart there, so the notes of both end up in a single VoiceData.

The writer avoids producing this. It only writes a voice into another voice's rests when every voice on the staff has its own <voice> text, so a score built through mx::api with two voices of the same name on a staff is still written with a <backup> between them and reads back as two voices. The open question is about files from elsewhere, and about whether mx should do more than it does now.

What needs to be decided:

• Whether one voice is the right reading. The XML gives nothing else to go on, so this may be the answer, in which case the work is to document it (for example on VoiceData::label) and close.
• If not, what in the file could tell the two voices apart. No such signal is known yet.

Related: #492, #425, #422