Add more specific examples to the docs

There are a few specific use cases where having access to `grape` in OAK would be particularly useful but requires some demonstration.
These should be in the docs.

For example:
* Similarity metrics
* Subgraphs
* Composing graphs by predicate type as preprocessing for graph ML

Start by locating the project documentation and the existing OAK/Grape examples. Add demonstrations for similarity metrics, subgraphs, and composing graphs by predicate type as preprocessing for graph ML. Done means each listed use case has a clear, working example in the docs.

Written by the indexing model from the issue text.