I wonder…
That lots of time has passed since the first proposal and there is no development interest in this project.
In general terms, we can consider that parallax mapping is a ‘shading hack’ it means that it works only on the “VIEW” side of things and is practically useless in the general context of Blender toolset.
I might consider that the Blender development community favors rock-solid implementations always instead of “clever” hacks, that would act as a foundation and as something that would be multi-useful in the broad side of things.
So actually what we consider what is most important thing here, is not actually to display a parallax occluded surface. But to have a proper and standard way to subdivide it efficiently.
Something like a new adaptive (takes into account: occlusion and view distance) and delauney based (that is able to create uniform triangles resulting in smooth surfaces) subdivision modifier. This is a much better goal to look for.