How do columnar databases differ from traditional row-oriented databases in terms of data storage organization, compression techniques, and query processing optimizations, and what are the specific use cases where columnar databases demonstrate superior performance over their row-based counterparts?