Is there any reason not to return the samples of rho, lambda, the sigmas, and the Us?
I think MCMC output users would like the empirical distribution of all these quantities, but I don't know if they're left out on purpose.
If not, I'll try to write this up in a new branch and submit a pull request. It shouldn't be hard.