Amazon Aurora MySQL provides:

Continuous, incremental backups to Amazon S3 with point-in-time recovery (PITR), allowing recovery to any point within the retention window (typically up to 35 days). This easily achieves an RPO of less than 5 hours, often down to seconds.

Support for database auditing parameters so audit logs can be retained and managed (for example, in database logs or external log services) to meet the 7-day audit requirement.

Auto scaling (via read replicas, Aurora Serverless v2, or capacity management) to handle variable read/write demand throughout the year with minimal operational overhead.

Why others are less suitable:

A: DynamoDB is NoSQL and does not directly satisfy typical relational database + SQL audit requirements; Streams also only retain 24 hours, not 7 days, and on-demand backups do not inherently give an RPO < 5 hours without frequent scheduling.

B: Amazon Redshift is a data warehouse, not a primary transactional application database.

C: Snapshots every 5 hours give an RPO of up to 5 hours, not less than 5 hours, and rely on discrete snapshot points rather than continuous PITR.