Document, Test, and Release the similarity metric implementation

[anjsimmo]

Currently there aren't any test cases (either formal or informal) for the similarity metric, other than some notebooks showing bugs I stumbled across when attempting to use it (there could be more).

Suggestions:

• Include notebooks demonstrating how to apply the similarity metric to EFDT trees, VFDT trees, and Scikit-learn trees (via conversion to Scott's Tree impementation), and check these make sense.
• Include examples from the paper How to Compare and Interpret Two Learnt Decision Trees from the Same Domain? that the metric is based on. The paper contains ambiguities and the results reported in the paper figures and text differs, but we should at least test which are consistent with our impementation and document our interpretation.
• Publically release the impementation of the similarity metric code (in its own repo) with a README so that others can use it (as it has a dependency on Scott's Tree impementation this will also need to be publically released). Consider putting it on https://paperswithcode.com/ as an implementation of the paper the similarity metric is based on.