Why can micro-partition metadata improve query performance?

1. It replaces the need for any compute
2. It forces all queries to scan every partition
3. It guarantees every predicate scans one partition
4. It helps eliminate partitions that cannot match predicates