There might be a solution of using automatically post processing the recorded videos from both cams.
First there are some questions to clarify:
1. What type of information you would like to give to the jury, an (one) image or a video
2. At which time the judges will take a look on the result, just after the jump or do they scan the jumpers afterwards one by one
3. For what do you need to capture the image at the time of the jump? To see the bib number or to see the take-off?
4. How do you start the recording, manually or using a trigger at a given moment?
5. What is the resolution of both cameras in use?
6. Do you use the same framerate on both cameras and how high is it?
As Joan already mentions, the mode of recording in principal does not influence the framerate but may be related to the system bandwidth.
It might be better to use the retroactive-mode, it gives a more stable video-streaming due to exclusively working in the computer memory without disk access interference. To my personal experience it is even more stable to use separate instances for each camera and not to use the dual recording mode in one single instance. Comparing both methods on my computer it has been shown that the total cpu-load rather is smaller if one Kinovea-instance controls one camera, compared to the dual recording mode.
A technical more simple solution would be to create a horizontal side-by-side video of the whole jump, using the 1. cam on the left side as the leading video that controls fps, video resolution and image quality, padding the 2. cam to the right side.