A pipeline failure should rerun only safe incomplete work without duplicating registrations or deployments. Which approach best fits?

1. Bind runs to versioned input data and capture configuration.
2. Define controlled retraining triggers and validation steps.
3. Use appropriately scoped workload identities and permissions for the stages.
4. Design idempotent steps and durable completion tracking.