[epic] optimise parameter search spaces for all models

[mastoffel]

Using available datasets.

Need to think about:

• whether to include more parameters as defaults (often autoemulate uses a subset)
• whether to extend/reduce the search space for each model

• optimise parameter search spaces for GaussianProcessSk #168
• optimise parameter search spaces for XGBoost #169
• optimise parameter search spaces for RBF
• optimise parameter search spaces for RandomForest
• optimise parameter search spaces for GradientBoosting
• optimise parameter search spaces for SupportVectorMachines
• NeuralNetTorch (MLP) (done by @bryanlimy)