Better syllabification 2: sonority-based syllabification

[msonderegger]

To be implemented after #217 : sonority-based syllabification.

An implementation that already exists is by David Mortensen:
https://github.com/dmort27/syllabiphon

... but it's not documented.

This makes use of the implementation of features and sonority from PanPhon, which is much better documented;
https://github.com/dmort27/panphon

Sonority-based syllabification will need to be language and phoneset-specific, to specify how different phones map to sonority levels.

There is no documentation for syllabiphon and some customization is required to get it to work for a concrete case. This repository includes examples of its use for Dutch and Chinese dialects:
https://github.com/cmu-llab/phonotactic-complexity-across-dialects

For our purposes: the first languages we'll want to support are English (and maybe Mandarin), and English is basically Dutch in terms of syllabification, so we should just be able to adapt what Shim/Chang/Mortensen have done here.

@mmcauliffe might have ideas here based on MFA.

This is all an enhancement, not to be implemented soon.