[Accessibility] - Hairpin in other voice is always read in all voices

NVDA always read a hairpin in global mode. Create a hairpin in voice 2, and navigate in voice 1 using NVDA. At the point the hairpin begins or ends, NVDA will read it. I tried this in Ms4, and it still exists, and it also doesn't read start and end position (bar and beat) of all spanners in Ms4, so an old bug plus a regression.