Support/use musical font files' x-offset for stems

[Jojo-Schmitz]

Resolves: musescore#25088
Resolves: https://musescore.org/en/node/369769#comment-1261591

Needs special treatment for Bravura, Emmentaler (particularly bad metadata apparenty), Gonville and Petaluma, these fonts' metadata seem to disargee with the stems being offset by half a stem width to the left for upstem notes/chords.

[Jojo-Schmitz]

This PR seems to prove that the CI action for vtests is basically useless :-(
It should have reported Emmentaler to fail as miserable as this

So now the 'special' offset is only applied to Emmentaler quarter notes

But more exceptions are needed, a holy mess...

[worldwideweary]

Here's to hoping you get this resolved...
even though it's not a pressing issue, it sounds like a big enough task to get all of it squared away.
That old school Valerio font would be kind of fun to have working well

[Jojo-Schmitz]

I'm quite happy with it now, was able to reduce the exceptions to Bravura, Emmentaler, Gonville and Petaluma, where for Emmentaler and Gonvilla we might be able to instead fix their metadata.
Remaining issue with Valerio is beams:

[worldwideweary]

Oh yeah that looks better. Two things I noticed: Check out the braces for the staves... on a grand staff. Its y-offset is imbalanced, and not sure the cause:

Also, those stems for the whole notes look like ever so slightly could be benefited from having the y-starting positions be some how reduced (nit pick) so that the width of the stem doesn't show up inside the "empty" part of the note head and blends instead with the head's sort of "stemmy" portions. But whatever, it's not a big deal:

[Jojo-Schmitz]

I wouldn't like to code Valerio exceptions, I'd rather restrict exeptions to builtin fonts that need them
But let's see whether y-offestting the stems for long notes by half a spatium.
Actually that seems to be the case, so I'll add it.

No good idea what's going wrong that that brace, it is only Valerio and only this brace, not braceSmall, braceLarge or braceLarger (which it'd fallback to be using Bravura's), Removing that brace's metadata doesn't seem to make any difference.
Special casing Valerio in bracket.cpp, line 133 helps:

diff --git a/libmscore/bracket.cpp b/libmscore/bracket.cpp
index 26f572fa48..aefd3ca0d7 100644
--- a/libmscore/bracket.cpp
+++ b/libmscore/bracket.cpp
@@ -130,7 +130,7 @@ void Bracket::setStaffSpan(int a, int b)

             if (v == 1)
                   _braceSymbol = SymId::braceSmall;
-            else if (v <= 2)
+            else if (v <= 2 && score()->styleSt(Sid::musicalSymbolFont) != "Valerio")
                   _braceSymbol = SymId::brace;
             else if (v <= 3)
                   _braceSymbol = SymId::braceLarge;

Or treat Valerio the same as Emmentaler and Gonville, i.e. draw the brace 'manually' rather than taking the font's glyph.
It'd then use braceLarge, and from Bravura, but I'm not sure I like that idea, see above. I'd rather have the font's Author to correct this in the font. I've asked in the Steinberg forum, https://forums.steinberg.net/t/valerio-a-16th-century-music-font/941064/13?u=jojo-schmitz

[Jojo-Schmitz]

The Valerio brace issue is resolved, in version 1.001 of that font

[Jojo-Schmitz]

There are more issues, but without working vtests I can only judge from the vtests of the port of this PR to upstream master, see musescore#25050 (comment)

[worldwideweary]

Are the vtests in 3.7 not as robust as the 4.x series? I wonder "whadda they got that I ain't got?" as related to this repo

[Jojo-Schmitz]

They apparently don't do anything usefull at all. on upstream master the method has changed quite drastical, here not at all or no much vs. the former working ones upstream 3.x

[worldwideweary]

Damn, and since layout is different on 4, can't just easily backport them...

[Jojo-Schmitz]

OK, taking a different approach now for the special treatments (changing the x-offsets for the 'broken' fonts), only for the affected glyphs (basically noteheadBlack and noteheadHalf) rather than the affected durations, either in code (for Bravura and Pertaluma) or via the fonts' metadata.json (for Emmentaler and Gonville). Similar for the defaults for noteheadDoubleHole (where all fonts seem to agree, except Legato)
Let's see whether that improves the failing vtests on master

[Jojo-Schmitz]

Fixed for Legato and also for AloisenU and AloisenGrooveU.

2 remaining issues (at least), in addition to the beaming issue with Valerio there's an issue with slurs, in all fonts (here shown with Valerio), and upstem as well as partly downstem: