[Haskell-cafe] why is rpar/rpar always faster (book Simon Marlow)?