What does partition evolution in Apache Iceberg allow without breaking existing queries?

Iceberg partition evolution allows the partition specification to change over time; new data uses the new partition scheme while old files keep their original layout; and queries are still correct because Iceberg tracks partition info per file. The incorrect options ("Changing the underlying file format from Parquet to ORC", "Merging two Iceberg tables into a single unified table", "Converting a managed Iceberg table to an external unmanaged table") are distractors that don't fully capture the concept described.