Why is this so hard to do, why couldn't a terrain tool allow for sculptris like tessellation and maybe a simple retopo tool to clean up the mesh/terrain.
Because, most commonly, terrain in 3D games are handled as a flat mesh to which a heightmap is applied. A single heightmap does not allow you to specify caves, overhangs, or other similar structures.
You can use multiple heightmaps, if your engine knows how to work with them. To simplify, one map bumps "up and down" and the next ones bumps "out/in". These maps are relatively trickier to generate or work with.
Depending on the game world, you can model the whole terrain, or you can model certain parts of it and "overlay" it with your heightmapped terrain.
Rarer techniques are voxel-based engines or procedural/higher-order surfaces based engines. But, those are not common.