Add Random Forrest Support for TreeSHAP-IQ

[mmschlk]

Similar to the already existing sklearn decision tree support, we need to also support sklearn random forests naturally.

For this a special mechanism in the conversion of tree-models to their edge representation is required.

Idea: RF average the predictions of the ensemble members. The same goes for their Shapley interactions. A list of EdgeTrees is probably right.

The implementation in the o.g. TreeSHAP-IQ should work.

Tasks

• Add RF converter
• Add tests