In this case the block is on the floor without any gap, like if the geometry was sliding across the ground. If you want to simulate a block that resembles a car, you could have a look at the Ahmed body.
As per the ground’s boundary condition, it mostly depends on what you want to correlate. If it’s an open road simulation, then you’ll definitely want to have a moving wall with a velocity equal to that of the vehicle but in the opposite direction —you take the vehicle as the reference frame—. On the other hand, if it is a wind tunnel you want to correlate to, then it will depend on the particularities of that wind tunnel. The more general wind tunnels have a fixed floor, while the ones built with automotive applications in mind will have a rolling belt that will go under the car between the tyres (like in the video), with the most advanced ones having a full rolling floor. So depending on the case, you should define your floor accordingly.
Regarding boundary layers, in the past I experienced that STL files with only one patch will generate no boundary layers. You’ll need to split the geometry into at least two patches. Also, I’m not sure whether it’s possible to get boundary layers or not with the automatic meshers; but in the parametric mesher I would have more cells in all directions. With the box dimensions and the number of cells you can know the base size. Then, knowing that each refinement level will halve the cell size, you can use that to define your refinements accordingly.