Fuzzy-match millions of rows in Databricks (2026)

When you fuzzy-match 10 million rows, you aren't "just comparing strings." A naïve dedupe implies...