On the possibility of ferromagnetism in carbon-doped anatase TiO2