This data is published by the business, not scraped.
Every record was submitted by an operator for machine consumption and carries
its own freshness contract. When a question cannot be answered from it, the
query API says so explicitly instead of guessing.