The Area Covered is an estimate according to the maximum extent of the Automatic Tie Points generated during step 1. Your observations are correct, this is not always the area of the orthomosaic. Depending on the image dataset, we expect the Area Covered to be larger than the area of the orthomosaic.
Here are the reasons: some of the Automatic Tie Points that you see in the rayCloud are discarded for the next steps of processing, as they are considered to be noise. We always discard 1% of the Automatic Tie Points in each cardinal direction (north, south, west, east). From our experience, the borders are usually noisy and we want to remove this noise.
The density of the points is very important. If your area has a dense point cloud in the borders, then you should not notice the discarded points (see my screenshot below):
The density of a point cloud depends on several parameters: quality of the image dataset, image overlap and the characteristics of the study area in terms of texture and geometry (more points will be generated in an urban landscape than a vegetated or sandy area).
For a more precise touch, I would recommend you to draw a processing area based on the densified point cloud.
Generally, we strongly recommend to trust the area of the rayCloud or the area of the orthomosaic.