fix(importer): read MusicXML jumps from <sound> inside <direction>

[Danielku15]

Note

AI-authored disclosure (alphatab-ai-authored-v1)

Portions of this content were authored by an AI agent. The agent has read
AGENTS.md and the human submitter accepts responsibility for
compliance with the rules in that document.

Issues

Fixes #2859

Proposed changes

MusicXML allows <sound> both directly in <measure> and inside <direction>. Notation
programs (e.g. MuseScore) write D.C., D.S., To Coda, Coda, Segno and Fine in the
<direction> form. The importer read the jump attributes (coda, tocoda, dacapo,
dalsegno, fine, segno) only from a measure-level <sound>. For a direction-level
<sound> it read only tempo, so these jumps never reached MasterBar.directions.

Both positions now share one mapping (_parseSoundDirections).

MusicXML keeps display (<direction-type>) separate from playback (<sound>). alphaTab has a
single Direction that both renders its own label and drives playback. Within one
<direction> the importer therefore treats <sound> as authoritative and the
direction-types as its visual counterpart. This is the same rule already used for
<sound tempo> vs. <metronome>.

<direction> content	Before	After
<words>To Coda</words> + <sound tocoda>	beat text only	JumpDaCoda, no duplicate text
<words>D.C. al Coda</words> + <sound dacapo>	beat text only	JumpDaCapo ("D.C."), no duplicate text
<words>Fine</words> + <sound fine>	beat text only	TargetFine, no duplicate text
<coda/> + <sound coda> / <segno/> + <sound segno>	Target*	unchanged
<coda/> + <sound tocoda> (coda symbol printed next to "To Coda")	bogus TargetCoda	JumpDaCoda only
<coda/> / <segno/> without jump attributes	Target*	unchanged
other words next to a jump (e.g. "Andante")	beat text	jump + beat text kept
<sound dacapo="no">	JumpDaCapo	no jump (dacapo is yes-no typed)

Words are dropped only if they are a known label of a jump present on that <sound>.
The match ignores case, dots and whitespace (incl. NBSP), and words split over several
<words> elements are joined first. The list covers MuseScore's default export labels and
the spelled-out forms. Double/numbered segno/coda labels are deliberately left out.

Not changed:

• No "al Coda" / "al Fine" inference. This follows the decision in MusicXML: <sound dacapo="yes"/> with a <sound tocoda> elsewhere plays to the end after the da capo instead of jumping to the coda #2860. The issue's
  repro now plays 1 2 3 1 2 3 4, the same as with measure-level <sound> today.
• Named jumps (segno2/coda2), time-only and forward-repeat are not handled.
• Other direction-level <sound> content (pan, <midi-instrument>, <swing>) is
  still ignored.
• Jump position: jumps stay at master-bar level. A jump written on the wrong side of a
  barline is not moved.

Checklist

• I consent that this change becomes part of alphaTab under its current or any future open source license
• This PR is linked to an accepted issue (see above)
• Changes are implemented
• New tests were added
• I have read AGENTS.md if an AI helped draft any part of this PR

AI authorship disclosure

• No AI agent authored any part of this PR (description, code, tests, or commit messages)
• An AI agent contributed to this PR. The AI-authored disclosure block
  (alphatab-ai-authored-v1) is present at the top of this body, and I have
  personally reviewed every change and can explain each one

Further details

• This is a breaking change
• This change will require update of the documentation/website