Why this was not considered one of the top priorities is beyond me. Displacement map baking is essential for game asset development which is one of the most popular use-cases of Blender. But this is a wider problem.
The baking workflow in Blender is ARCHAIC. They’ve implemented the Principled Shader that is currently the most widely used and you still cannot bake Principled Shader values to maps without creating a node setup just for that purpose and baking values one by one using simple shader color value. So it’s easily doable. You just have to forward some of the values as black to white scale. Why is this taking years to implement? Also, currently there are people that successfully emulate parallax occlusion mapping in Blender using Eevee and no one looked at that and said “that’s an obvious choice for Eevee functionality, isn’t it?”.
The state in which Blender 2.8 Release Candidate is now is basically crippled. Sure - the progress overall is big, but there is some unforgivable regression in core functionality. I think releasing 2.8 in the current state is a big mistake and may attract some new users to immediately push them away due to the utterly broken baking workflow.