Re: [3/3] d3d9: account for another different behavior in the viewport test