Add an example with Gaussian copulas

[rlouf]

It is fairly simple to generate samples from a multivariate distributions with arbitrary marginals using Gaussian copulas:

1. Generate a vector of samples from a multivariate normal distribution;
2. Apply the cdf of the standard normal distribution to each element of these vectors;
3. Apply the inverse cdf of the marginal distributions to the corresponding elements.

Yet some PPLs wrap the whole process in custom distributions (Numpyro, TFP). I thus suggest we illustrate Aesara/AePPL's modeling capabilities on a copula example.