Codespaces idle timeout and storage cost management

GitHub Codespaces allows organizations to manage compute costs by configuring idle timeout settings, which automatically suspend environments after a period of inactivity. While compute charges stop once the instance is suspended, storage costs for the persistent disk continue to accrue regardless of the machine's power state.

For low-traffic workloads, there is a need to balance aggressive compute timeouts with the long-term cost of retaining stopped codespaces. Organization-level policies can enforce maximum machine sizes and timeout limits, but the interaction between the retention period and the idle timeout determines the total cost per user.

What is the precise behavior of the retention period when a codespace is automatically stopped by the idle timeout? Can storage costs be minimized without manually deleting the environment after every session?