core: fix async task-less execution of remove-image

RemoveImage may lack async tasks when:
1. The image doesn't exist
2. There's a failure during delete image

With these changes, the image would be removed from the database
(rather than staying in LOCKED state) and the command would finish
successfully in the first case, and would end with failure with
the image switching to ILLEGAL state in the second case.

Signed-off-by: Arik Hadas <ahadas@redhat.com>

backend/manager/modules/bll/src/main/java/org/ovirt/engine/core/bll/storage/disk/image/RemoveImageCommand.java

@@ -138,12 +138,14 @@ protected void executeCommand() {
                 if (e.getErrorCode() == EngineError.ImageDoesNotExistInDomainError) {
                     log.info("Disk '{}' doesn't exist on storage domain '{}', rolling forward",
                             getDiskImage().getId(), getStorageDomainId());
+                    endSuccessfully();
                 } else if (e.getErrorCode() == EngineError.ImageDeleteError && isImageRemovedFromStorage()) {
                     // VDSM renames the image before deleting it, so technically the image doesn't exist after renaming,
                     // but the actual delete can still fail with ImageDeleteError.
                     // In this case, Engine has to check whether image still exists on the storage or not.
-                    log.info("Disk '{}' was deleted from storage domain '{}'", getDiskImage().getId(),
-                            getStorageDomainId());
+                    log.info("Failed to delete Disk '{}' from storage domain '{}', changing to illegal",
+                            getDiskImage().getId(), getStorageDomainId());
+                    endWithFailure();
                 } else {
                     throw e;
                 }