what number of class ?

Here AWS defines HA, and uses the word cluster - AWS has several methods for achieving HA through both approaches, such as through a scalable, load balanced cluster or assuming an active–standby pair. - https://docs.aws.amazon.com/whitepapers/latest/real-time-communication-on-aws/high-availability-and-scalability-on-aws.html

Option C proposes launching two EC2 instances in different AWS Regions and setting up database replication, with failover to a second Region. While this solution does provide geographic redundancy, it may introduce higher latency due to cross-region communication and data replication. Additionally, failover to a different Region typically involves more complex configurations and longer recovery times compared to failover within the same Region. While Option C may offer a level of redundancy, it might not provide the same level of responsiveness and automatic failover capabilities as Option A, which leverages Availability Zones within the same Region. In scenarios where low latency and rapid failover are critical, Option A is often preferred. However, if geographic redundancy is a top priority and the potential trade-offs in latency and failover time are acceptable, Option C could still be a viable solution.

but for C you need communication between the two VPC, which increase the complexity. With a should be enough for HA

The question doesn't ask which option is the most HA. It asks what meets the requirements.

Design for region failure? may as well design for AWS failure and put replica in GCP and Azure :v