Tell a negative octave from a flat in removeRedundantPitches

[float3]

Pitch.nameWithOctave spells both B-flat in octave 1 and B-natural in octave -1 as 'B-1', because the '-' is the flat sign and the minus sign both. removeRedundantPitches compared pitches by that string, so the second read as a repeat of the first and was thrown away.

Revised after review: _removePitchByRedundantAttribute now has the special knowledge of 'nameWithOctave' itself and compares by name and octave in its place. Its signature is unchanged, no tuple form is exposed, and both call sites and the pitch-class and pitch-name reductions are exactly as they were.

While rewriting it I noticed the earlier version had a second problem, so this one also pins it: the octave is compared together with octaveIsImplicit, so a pitch given no octave stays distinct from one placed in the default octave. nameWithOctave already made that distinction ('C' against 'C4'), and since v11 Pitch.octave returns 4 for both, comparing the octave alone would have quietly merged them. There is a doctest and a test case for it.

The docstring documented the old behaviour as a known bug ("doesn't seem a bug worth squashing at this moment"); it now documents what the method does, and test_chord carries a regression test.

[coveralls]

coverage: 93.348%. remained the same — float3:fix-redundant-pitches-negative-octave into cuthbertLab:master

[mscuthbert]

I would rather have removeByRedunantPitches have special knowledge of 'nameWithOctave' and substitute 'nameWithOctave' with a special routine (either as a tuple or not) -- make the equivalence not something users have to think about.

Based on that, do not demonstrate ('name', 'octave') as the usage of a tuple in the docs -- I was trying to figure out some other usage of the tuple form -- but i can't think of one -- everything i tried ('pitchClass', 'octave') had an equivalent single attribute like ps -- so unless there's another good use for tuple form, do not expose that. Complexity reduction is a direction we're trying to go with -- this is the opposite. :-) Thanks!