Hello vcc,
Thank you for providing the project file.
The issue seems to be related to the inaccurate camera interior orientation parameters estimation. Which is likely caused by the mission plan that doesn't include 180-degree rotation of the camera on the neighboring flight lines.
I've tried to re-align the cameras after enabling Fix Calibration flag and the were no splitting of the areas in the result, however, the bowl effect could be still observed. The optimization in the version 1.5.0 enabling photo-invariant calibration parameters, however, seems to fix the problem. I have sent the final project to you via e-mail.