[Interface to MLJ] Interface to supervised MLJ models

[RaunoArike]

MLJ is a machine learning framework that provides a large suite of common machine learning models, both supervised and unsupervised. This package can't currently be interfaced to MLJ due to differentiability issues: to be able to use any of our counterfactual generators to explain MLJ models, those models need to be differentiable with respect to features.

To solve this, we will take the following steps:

• [x] Implement a basic interface to MLJ (an AbstractFittedModel for MLJ.Supervised)
• [x] Investigate which models are compatible with the package, taking the steps outlined in issue #174
• [x] Test all implemented functionality
• [x] Document the entire process

[RaunoArike]

• Basic interface: #206
• Implementation of EvoTrees from MLJ: #206
• A report detailing our investigation into MLJ models: #206
• Tests and documentation for EvoTrees: #206
• Implementation of the DecisionTree and RandomForest models from MLJ: #195
• The Feature Tweak generator for generating counterfactuals for DecisionTrees and RandomForests: #195
• Tests and documentation for DecisionTrees, RandomForests and Feature Tweak: #195
• The Growing Spheres generator for generating counterfactuals for EvoTrees, DecisionTrees and RandomForests: #243
• Tests and documentation for the Growing Spheres generator: #243