Allow multivariate transformations of parameters in mcmc plots

[jgabry]

Currently the transformations argument to the mcmc_* functions only handles univariate transformations (i.e., transformations of scalar parameters).

[jgabry]

This is somewhat related to the RStan issue stan-dev/rstan#517. Right now it’s a pain to plot unconstrained parameters in bayesplot unless it’s a univariate transformation. But if we can make it convenient to get the unconstrained draws from rstan then perhaps that’s mitigated since those draws could be passed to bayesplot and avoid needing the transformations argument.