A good starting point may be Intel's white paper on connecting together multiple D415 cameras with the hardware sync function.
The paper says about collecting synchronous frames:
if no HW sync is applied, frames will be available asynchronously from each camera. If the “capacity” is set to the minimal number (1 for depth-only, and 2 for depth+color), then only the most recent frames will be available at any time, so frames from different cameras can never misalign by more than 1 frame.
If even better alignment is required, then the HW sync should be applied. In this case all cameras will fire and capture frames at essentially identical times. Since the calls are blocking from each camera, it should simply be a matter of collecting all frames and displaying them.