How are people doing metadata management in large Azure data architectures? I\'m particularly interested in how people are managing and versioning schemas for schema-on-read da