Retracted autism study an 'elaborate fraud,' British journal finds