Looks good. For testing just compare output as gotten from my package with old LP version. I haven't run all regression on the new
persian.ly (that works with current LP version) yet.
It's actually quite simple: everything is the same except for 2 new accidentals Koron and Sori which affect pitch as documented. The -v suffix affects only MIDI and could be dropped, was more of an experiment.
It seemed I could not add these extra accidentals (and their pitch in MIDI) but had to redefine all accidentals.
Ugly but it has worked fine for over a decade.
Proper integration with LP would be better but hasn't happened.
Anyways it's now working fine with current version so I'm happy.
Thanks,
Kees