Priors for covariance matrix for multivariate normal

[bburan]

@aseyboldt

I think a reasonable workaround would be to use the LKJCholeskyCov prior you already defined, but add a second observed NvNormal to include the data from the paper.

That seems like a good solution, thanks!