[Request] Skew-t, multivariate skew-normal, multivariate skew-t

The skew-normal distribution is available in Stan.
The skew distributions are generally of the form:
2*f(x; theta)*F(z_x * alpha; theta_z)
where f is a pdf and F is a cdf.

Thus, the [log] skew-t can be implemented as:
log(2) + student_t_lpdf(x | nu, xi, scale) + student_t_lcdf(alpha * (x - xi)/scale | nu, 0, 1)

This is likely inefficient. Moreover, the multivariate skew-normal and skew-t are available. See:
http://onlinelibrary.wiley.com/doi/10.1111/1467-9868.00391/abstract multivariate skew-t
https://arxiv.org/abs/0911.2093 multivariate skew-normal
http://www.sciencedirect.com/science/article/pii/S0047259X11001126?via%3Dihub centered parameterizations (and benefits of doing so) for the skew-normal and skew-t (you all currently use direct parameterization, which I don't think is an issue for bayes, but has some odd log-lik properties at around alpha=0, and thus could be an issue for optimization and the hessian).

No implementation file or test is identified. Start by locating the existing skew-normal distribution in Stan Math, then review the cited skew-t, multivariate skew-normal, and centered-parameterization references. Done means the requested univariate and multivariate distributions are implemented with appropriate parameterizations and validated behavior.