ci: disable RDS automated backups in the temporary Terraform test roots

[bobbyiliev]

The two temporary AWS roots inherit the database module's backup_retention_period default of 7 days, so each run's throwaway RDS instance keeps a week of automated backups that its teardown then has to delete. Nightly 18272's terraform destroy was still working on the RDS instance after 55 minutes against a 90 minute step budget, having taken 1m51s on 2026-08-24, and the step timed out with the cluster, VPC and KMS key still up. skip_final_snapshot was already true, so the final snapshot was not the cause. Setting the retention to 0 disables automated backups, which a test database that lives for under an hour has no use for, and drops the backup storage they were accruing.

aws-persistent is deliberately left alone: it is long-lived, so the trade-off there is not the same.

Test plan

terraform validate passes on aws-temporary and terraform fmt -check -recursive is clean. Whether it actually shortens the teardown shows up in the two AWS nightlies' destroy timings, which is also the only place the previous 55 minute run was visible.

🤖 Generated with Claude Code