What role do columnar storage formats, such as Apache Parquet and Apache ORC, play in optimizing table storage and query performance for analytical workloads, particularly in big data and data warehousing scenarios where efficient compression and vectorized processing are crucial?