Author new documentation

[trevorbaca]

• Document LilyPond text script alignment; figure out why abjad.tweak(abjad.Right).self_alignment_X appears to do nothing
• Write a comparative overview of LilyPond spanners
• Write (and web publish) an explanation of exactly how score segments work; which variables are implemented to work across segment boundaries; which features in the build system are designed to handle segment-crossing settings; etc
• Write an explanation of how the Bača "externalization" scheme works for part extraction from large scores; document LilyPond tag limitations in that a single part can't move between staves). This is probably part of a larger write-up that will document exactly how the setup of large scores works comprehensively between Bača classes and LilyPond features (tags, keep-alive, new context instantiation)