Improve Image delete actions

[rsmontero]

Description There are some operations that may leave images in the datastore. Although this situations are logged, it is not properly signaled, and it may lead to stale images.

The idea is to leave the images in DELETE state (and maybe ERROR_DELETE / LOCK_DELETE) to be able to retry the delete operation; or manually check the status of the associated image in the datastore.

This feature needs a recover procedure (e.g being able to retry oneimage delete on the states, and probably force operation to delete the db object).

Points where this situations may occur:

• A delete operation itself fails
• A CP operation succeed but the image has been marked to be delete e.g. leave the image in LOCK_DELETE to tigger rm and track result of the subsequent delete operation.

Use case Do not leave stale images in the datastore. Recover from delete errors.

Interface Changes Probably New state images

Additional Context

Progress Status

• [ ] Code committed
• [ ] Testing - QA
• [ ] Documentation (Release notes - resolved issues, compatibility, known issues)