Create HMC notebook

[FarmHJ]

Closes #38.

The cone problem seems to be a very difficult problem. Do we want to keep this problem for the testing notebook?

[MichaelClerx]

I don't know enough about HMC @FarmHJ , so you'll have to rely on @ben18785 for this one!

Small comments / (possibly bad) ideas:

• Should we show a small bit of single-chain exploration on e.g. a parabolic surface? Would like to see the "hamiltonian" bit in action
• Is there something that can be done with hyperparameters to improve cone performance?
• One of the headers says multimodel instead of multimodal