Add a new detection metric: string similarity

[daanraman]

Add new detection method: similarity between strings. Could be interesting in specific use cases to identify terms that are entirely different from others, in cases where we would expect a lot of similarity.