To align with modular changes elsewhere in KG-Hub, we should consider storing the similarity tables produced here somewhere other than KG-Hub S3: * GitHub release (Jenkins build would need to create draft release) * Zenodo * HuggingFace Data