You're right, the colour of materials is much more complex than just the available atomic transitions. The electronic states are changed and spread out by being bonded to other atoms and you get bulk collective properties like polarizability. The relativistic contraction mentioned in the article is still relevant for the electrons' energy levels in bulk gold (a big lump much larger than the wavelength of visible light) and so neglecting it would give the wrong colour for bulk gold.

Structures up to a few hundred nanometers in size could also conceivably affect the colour. This is kind of like how the atomic-level description of a radio antenna doesn't really matter, more its bulk properties like conductivity. The gold nanoparticle effect is due to Mie scattering (http://en.wikipedia.org/wiki/Mie_Scattering), which is the scattering of light off dielectric spheres approximately the size of the wavelength of light. This effect isn't relevant if you just have a brick of gold.