Make (some) notebooks executable without needing to sample

#239 will set up badges to open the example notebooks on github, binder or colab. Colab should be able to run most if not all notebooks without too much trouble, but binder is more restrictive when it comes to available resources. It would be amazing if we could get in an automated or semi automated way publish the inferencedatas of some notebooks (maybe the more popular ones) so that they can be executed and people can explore the model and the results without having to sample.

Bonus points. Right now this is useful but doing so looses part of the attractive of the notebooks: playing with the models, see what and how affects sampling... If/Once we get starting point, mass matrix... into the inferencedata and make available a way to use those, then by having those inferencedata available, users will be able to run and play around with sampling without needing to run tuning.

Start by reviewing issue #239 and the repository's example notebooks to understand how badges and notebook execution are handled. Define which notebooks should publish inference data and what automated or semi-automated process would make them executable without sampling; the bonus scope includes making starting points and mass matrices available for later sampling.