Veblen’s theory of the leisure class revisited: implications for optimal income taxation