Updated version following reviews

Quite a massive update following a first round of reviews, in particular in what concerns the technicalities of the Bayesian models, including the use of an alternative set of priors, the inclusion of a second global language phylogeny, and the inclusion of a 2D Gaussian Processes to better capture contact.
Nevertheless, the results are largely unchanged.

Version acompanying the initial submission

v1.0

Update README.md