Cannot update Elasticache Replication Group in cluster mode with 0 read replicas

### Name of the resource

AWS::ElastiCache::ReplicationGroup

### Resource Name

_No response_

### Issue Description

We have a replication group with replicasPerNodeGroup: 0 and automaticFailoverEnabled: true, but the replication group has Cluster mode enabled.

This works fine when creating the cluster, but when we try to update a property of the cache group (we are trying to update the security group), we get the error:

`Replication group must have at least one read replica to enable autofailover.`.

We cannot turn automaticFailoverEnabled to false because it errors out with `Turning off automatic failover for replication group with cluster mode enabled is not supported.`

Updating replicasPerNodeGroup creates a new cluster, which is not ideal either.

### Expected Behavior

Security group of the cluster is updated without interruption.

### Observed Behavior

Cloudformation throws the `Replication group must have at least one read replica to enable autofailover.`. error.

### Test Cases

Create a clustered Redis cache with 0 read replicas.

Try to update the security group with VpcSecurityGroupIds.

Try to update any field [marked with no interruption](https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/aws-properties-elasticache-cache-cluster.htm).

### Other Details

_No response_

Reproduce the issue with an AWS::ElastiCache::ReplicationGroup configured for cluster mode, replicasPerNodeGroup: 0, and automaticFailoverEnabled: true. Follow the listed test cases by updating VpcSecurityGroupIds and other no-interruption fields; done means the security group and supported properties update without interruption or the conflicting service limitation is documented.

Written by the indexing model from the issue text.