Thanks for clarifying this. It looks like the latest 4.3 spec has been corrected in one place, but there's still a bug further on in the spec.
At first it says:
The number of isolines generated is derived from the first outer tessellation level; the number of segments in each isoline is derived from the second outer tessellation level.
But then says:
The u = 0 and u = 1 edges of the rectangle are subdivided according to the second outer tessellation level...Each of the n lines is then subdivided according to the first outer tessellationlevel.