Do not assume that util.skopeo_copy can't fail

If the disk is full (projectatomic#1146),
atomic will continue and show errors later, which is puzzling and hard to debug.

Author: mscherer

Atomic/backends/_docker.py

@@ -343,9 +343,8 @@ def pull_image(self, image, remote_image_obj, **kwargs):
         trust = Trust()
         trust.discover_sigstore(fq_name)
         util.write_out("Pulling {} ...".format(fq_name))
-        util.skopeo_copy("docker://{}".format(fq_name), image, debug=debug, insecure=insecure,
+        return util.skopeo_copy("docker://{}".format(fq_name), image, debug=debug, insecure=insecure,
                          policy_filename=trust.policy_filename, src_creds=src_creds)
-        return 0
 
     def delete_container(self, container, force=False):
         return self.d.remove_container(container, force=force)

Atomic/run.py

@@ -108,7 +108,9 @@ def run(self):
             if self.args.display:
                 return 0
             try:
-                db.pull_image(self.image, remote_image_obj)
+                err = db.pull_image(self.image, remote_image_obj)
+                if err:
+                    raise ValueError("Unable to pull image {}".format(self.image))
                 img_object = db.has_image(self.image)
             except RegistryInspectError:
                 raise ValueError("Unable to find image {}".format(self.image))