Merge DyHPO

[karibbov]

Merge work from a private repo on DyHPO, DPL surrogates, and related acquisition functions. Version without refactoring. All the further changes should be made under this branch.

Added Surrogates:

• DeepGP: Deep Gaussian Surrogate original implementation, paper
• DPL: Deep Power Law ensemble surrogate original implementation, paper

Added Acquisition functions:

• MFEI: Multi-Fidelity Expected Improvement
• MFPI: Multi-Fidelity Probability of Improvement
• MFUCB: Multi-Fidelity Upper Confidence Bound
• MF{EI, UCB, PI}AtMax: versions of each where the incumbent value is the best value at the highest fidelity
• MF{EI, UCB, PI}Dyna: versions of each where the incumbent value is the best global value and model extrapolate till the highest seen fidelity + 1
• MF(EI, PI)Random: versions of each with extrapolation at random horizons with random thresholds

Added Optimizers:

• MFEIBO: DyHPO, DPL and many similar optimizers can be obtained by combining corresponding surrogates with acquisition functions above

[Neeratyoy]

Could we squash commits?

[Neeratyoy]

Also we should probably add some basic unit tests to check for freeze-thaw sampler and MF-EI. As a second stage, dyhpo and dpl surrogate fitting (okay if this is not part of this PR).