Tolerate transient poll errors during clone wait

awaitMirrorReady previously aborted the whole wait on any GetMirror error, so a brief network/API glitch during a long initial clone failed mirror create even though the clone was still progressing. Retry on the poll interval, giving up only after maxConsecutivePollErrors consecutive failures (resets on success) or when the context ends — a persistent error (deleted mirror, revoked auth) still surfaces instead of spinning to the deadline. (Cursor)