Implement Baselines

[TobyBoyne]

We want to compare against a range of baselines.

Currently implemented

By using BoFire/BoTorch, we already have access to:

• GPs
• Random sampling
• Entmoot

To be implemented

• Prior bark (to verify that the posterior MCMC is even worth it)
• LeafGP (train a fixed GP. This mostly already exists, we just need to set up the BoFire interface)
• SMAC (this kind of exists in a dead repo, @TobyBoyne needs to clean up that code)
• BART (this code also exists, needs to be cleaned up. Also note that no optimization over tree ensembles exists AFAIK, so we need to just evaluate on a grid)
• Probabilistic Reparameterization (not necessary, this is just a fast way of optimizing the AF for mixed spaces - shouldn't be necessary, hopefully)