ALTO support for encoding OCR segmentation ambiguity

[artunit]

At the 2019-05-07 ALTO face-to-face Board Meeting there was support for identifying a a standard and interoperable lattice structure for encoding OCR uncertainty and alternative hypotheses, and the discussion since then has largely been attached to issue 57 - Allow a String to contain alternative Glyph segmentation hypotheses. Significant progress has been made, including an extensive proposal from @bertsky, and with momentum from the 2019-11-03 f2f meeting, it makes sense to move the discussion to a new issue. Further updates will be tracked here.