Currently, the feature-extractor only takes in a single database and collection argument. However, some .test files reference multiple databases and collections, so we should handle that during feature extraction.