Most BRDFs that I have seen get darker as they get rougher like the article mentions. There are some that take into account the interreflections of the theoretical facets that make the distribution of normals that become the BRDF (mentioned in the link also), and they do better but I'm not sure even they pass this test straight up.
Most of what is actually being used out there loses energy. Sometimes people have used a lookup table on top of the BRDF to compensate.
It doesn't pass the white furnace test.
That made me realize I didn't understand much of the math I implemented. Random multiplications by pi and 2 as the author stated didn't work out.