This is what you get when you release software in 2006 and after almost 6 years still don't have a proper manual explaining these things.
Leaving us experienced users to keep it going by repeatingly answering the same questions over and over again. Where did they all go, except for a couple/few?
Elipsis1, don't feel offended by that. You're question is upright and although there is a search it isn't always easy to find what you're specifically looking for especially if you don't know what's wrong which in turns has to do with the very first thing I mentioned.
Anyway, here we go...
The information on what to put where is very mixed and unclear now and then.
What Planetside tried to do with creating a terrain group and a shaders group is to create a common workflow or consensus on how to set up a scene.
Summarized: create your terrain and displacements in the terrain group and texture it in the shaders group. That's the basic idea.
However, this does not mean that if you alter your network in the shaders group to only process colour information that the displacement data will be preserved.
That's what happened here: you mixed your displaced terrain with texturing and mixed it by colour (mix colour 01).
The output of the mix colour 01 is colour and not displacement. So you end up with a flat surface.
See attached file for a possible solution.
What I did is I created a surface layer and hooked that up in the main network.
So from the compute terrain to base colours to surface layer = terrain displacement + colours from 'base colours' as data input into the surfacelayer.
The output of the mix colour 01 is then connected to the colour input of the surface layer. The diffuse colour of the surface shader is set to 1, which means that the colour input will be multiplied with 1, so un-affected (0.5 = half time the brightness of the input).
The simplest solution is to replace the 'mix colour 01' with a 'merge shader'. The merge shader merges both displacement and colour data by default.
I attached the previous example to try to explain different data types/flows/usage.
I hope it's a bit more clear now that the data flows from top to bottom to the planet node and that all data should remain intact.
The node group paradigm can be useful, but basically if you know which data goes where and what node accepts/outputs which data then you won't need any of these paradigms.
Good luck!
Cheers,
Martin