Setup markdown and test PPL comparisons

[MansMeg]

We should write down how to make comparisons between PPLs so that we test that the posteriord is the same (up to a proportionality constant).

Preliminary idea is to use log densities evaluated at ref draws (or random draws) and the gradients of the log density. We will store this in posteriordb?

[MansMeg]

• See if we can get PyMC and theoretical model to agree
• Add comparison also with stan log density
• Think about potential hypothesis test to compare two empirical distributions
• Jakob write up a summary in markdown