Incomplete sleep_study case study

[aloctavodia]

• There are a few sub-sections that explicitly use an LKJ for correlated priors. Allow non-independent random effects bambinos/bambi#702
• Exgaussian and lognormal families are not implemented in Bambi. Should we implement them in Bambi or define them in the case study

[aloctavodia]

@tomicapretto do you have a suggestion?

[tomicapretto]

The exgaussian and lognormal families could be implemented in Bambi with relative ease. For the LKJ prior, that will be harder, since the construction of the priors is independent right now. Anyway, I'm keeping this in mind for the "new bambi" backend, so it becomes possible.