New CloudWatch alarms with ok_actions email once each on creation

When Terraform/OpenTofu creates CloudWatch alarms that have ok_actions pointing at an SNS email topic, every alarm sends one email as it moves from INSUFFICIENT_DATA to OK (or ALARM) a few minutes after apply. With 8 alarms the user gets a burst of ~8 emails and may think something broke; warn them beforehand. Any alarm that lands in ALARM on creation is a real pre-existing condition (here RDS FreeableMemory under 100 MB on a db.t4g.micro), not noise. Also note ECS service health checks often live under a versioned path like /v1/health, so a 404 on /health is not an outage.