Could you elucidate the scalability features inherent to MariaDB, including its support for multi-master replication, sharding methodologies, and distributed transaction processing, and how these capabilities contribute to handling large-scale data workloads ?