fix #78636: insert before frame fails

[MarcSabatella]

If you select a frame in a score that does not have linked parts, then try to insert either a measure or a frame before it, the measure/frame is actually appended to the end ather than being inserted before the selected frame. That's because searchMeasureBase() was returning nullptr in this case. Fixed to return the original frame.

BTW, in the case of inserting measures, this bug was also "corrupting" any spanners present after the insertion point, because undoInsertTime() was still inserting the time at the proper place even though the measures themselves were being appended to the end.

[MarcSabatella]

Yikes, allowing the insert before a frame to succeed leads to a crash if you try to insert a measure before a frame. I guess that's why it was disabled. Still, hopefully another easy fix.

[MarcSabatella]

Crash fixed; there was an ill-advised cast from MeasureBase to Measure. Changed to use tick2measure() instead.

[MarcSabatella]

I thought about that, but based on my understanding of what an FBOX was intended for, I am not sure it makes sense from a user perspective to select one before inserting a measure, any more than it makes sense to select a fret diagram, or a dynamic marking or any other random element. You'd select the measure the FBOX appears to be attached to (even if internally, it isn't implemented that way - still not clear).

[Jojo-Schmitz]

OK, makes sense ;-)

[MarcSabatella]

We might want to consider this for 2.0.3. Not so much on account of the fact that frames are appended instead of inserted (a nuisance, but not a serious problem), but because of the bad effect on spanners. The fix itself should be pretty safe I think.

[Jojo-Schmitz]

I find the nuisance bad enough and as long as it doesn't break compatibility I don't see a real reason not to include it