Fix managed Helm upgrades wedged by migration Job S3 boot crash

Managed Tuist installs upgrading via this Helm chart could hang on helm upgrade --wait because the pre-upgrade migration Job crashed at boot. The chart gated the TUIST_S3_* env block behind not managedSecrets, and a recent change (#11460) removed the encrypted secrets blob that previously supplied the endpoint, so Environment.s3_endpoint/1 returned nil and runtime.exs hit URI.parse(nil) before any migration ran. The 0.12.1 chart hoists the S3 boot config (bucket name, endpoint, region, virtual host, bucket-as-host) on the migration Job out of that guard, mirroring the server Deployment, so managed upgrades now boot and migrate successfully. S3 credential refs stay gated since the migration Job does not call S3. Self-hosted (managedSecrets: false) installs are unchanged.