See https://musescore.org/en/node/289548

This is a regression caused by my continuous view performance optimization. Specifically, the change I made to articulation layout in order to fix an issue that was specific to continuous view. But I didn't actually need all the code I actually added to fix that issue. One bit I added - thinking it would be a nice little code simplification that would also help facilitate some future improvement - turns out to have caused this regression. I had moved the articulation layout slightly earlier, but it turns out this doesn't work because we haven't yet assigned the direction (up/down) and thus haven't necessarily chosen the proper symId, which throws off the bbox. Thus, the initial layout is bad, but later on we do calculate direction and assign the correct symID and so subsequent layouts are OK. if lter the direction is fliped, then the same glitch happens and fixes itself again.

This PR fixes the issue by reverting that portion of my continuous view change - moving articulation layout back to Chord::layoutArticulations() instead of Chord::layoutPitched(). As I said, this change wasn't needed anyhow, and performance is the same either way - even though there are multiple call sites now, each articulation still gets laid out only once (and it's a very cheap operation anyhow).