Enabled Optimized Iceberg Vacuum by default in Spark for premium tier clusters.

Enabled Optimized Iceberg Vacuum by default in Spark for premium tier clusters. When enabled, Dataproc will use a shuffle-based approach to rewrite data files, which can improve performance for MERGE queries that modify a large number of files. To disable this feature, set spark.dataproc.sql.optimizer.lowShuffleMerge.enabled to false.