humdrum: remove unneeded Regexes, fix partial-beam direction

[mscuthbert]

Third and last PR broken out from #1706

Reduce the number of regex searches needed to parse a humdrum scope -- Chopin moves from 6000 to 3400. Simplify the code and make it more elegant.

But barely moves on performance. hdStringToNote speeds up by 5% 8.59 µs/call to 8.14. Full parse speeds up by about 1%

Performance/cleanup, with identical output:

• accidentals: count '#'/'-' directly (the count is the alteration) and build a pitch.Accidental, instead of two re.search calls;
• rational durations: only run the \d+%\d+ search when '%' is present;
• pitch: only search when the event is not a rest (walrus in the elif).

Also, from ideas in PR #1706 (Alexander Morgan): comment out the phrase-mark, slur, and arpeggiation blocks that only ran pass, and rename the unused beaming loop variable from i to _.

AI-assisted (Claude). Read closely by Myke -- co-author Alexander Morgan

[coveralls]

coverage: 93.103% (+0.002%) from 93.101% — humdrum-kern-regex into master

[alexandermorgan]

Even if you don't want to factor out duration evaluation to its own function, I would still recommend doing it before anything else in this hdStringToNote function. The reason is that, e.g. for the case of rests here on 2366, you're still creating an initial duration that will just get replaced with the actual duration when you calculate it at the end of the function. So that's still 2 duration objects per rest which might account for the speedup being a little less than what I measured (10-12%). But that will increase the diff size quite a bit.

[alexandermorgan]

I had added these 4 tests and I would recommend keeping them to cover the new functionality. I had to add KernSpine.parse() because the new version of the KernSpine class got rid of the __init__ constructor, so some of the class properties were only available if I called .parse() first.
If you don't like calling .parse() there's probably a different way to run these tests, but I would definitely keep the tests in some form.

    def testGraceNote(self):
        ks = KernSpine()
        ks.parse()
        # noinspection SpellCheckingInspection
        a = ks.processNoteEvent('4Cq')
        self.assertEqual(a.duration.type, 'quarter')
        self.assertEqual(a.duration.slash, True)

    def testGraceNote2(self):
        ks = KernSpine()
        ks.parse()
        # noinspection SpellCheckingInspection
        a = ks.processNoteEvent('16Cqq')
        self.assertEqual(a.duration.type, '16th')
        self.assertEqual(a.duration.slash, False)

    def testChord(self):
        ks = KernSpine()
        ks.parse()
        # noinspection SpellCheckingInspection
        c = ks.processChordEvent('8C 8E')
        self.assertEqual(len(c.notes), 2)
        self.assertEqual(c.notes[0].duration, c.duration)

    def testChord2(self):
        # noinspection SpellCheckingInspection
        ks = KernSpine()
        ks.parse()
        c = ks.processChordEvent('8C E')
        self.assertEqual(c.notes[0].duration, c.notes[1].duration)