Minor changes for "official" release

[tresoldi]

This PR follows discussions here, here and here, plus a couple via email.

I did not change what is needed for a new release (like bumping the version in setup.py), but I already ran all the clts stuff which are needed for regenerating the data.

In detail, the changes are:

• Removed the conflicting diacritics for rounded/more-rounded and unrounded/less-rounded; only those related to the rounding feature are now used, as per IPA, further changes in terms of compression, protrusion, labialization of consonants, etc. are postponed to the next version

• Added a normalization for ɝ

• Implemented @LinguList solution for the bug of tones on vowels not properly handled (#96)

• Added all the linguolabials that someone might require, which includes all of those on Wikipedia (even the one from Coatlán Zapotec described as "mimesis for a pig drinking water")

• Added the code by @LinguList that allows to generate dataset statistics from the command-line (clts dstats)

[LinguList]

Thanks, I'd say this is okay, feel free to merge.