This is actually a pretty interesting question, as latency and FPS can be tricky concepts at first. (I'm a writer by trade, so they were for me too.) It sounds like you're confusing latency with frames per second. FPS imposes some latency by its very nature, but they're not the same thing.
The game thread should run at 120 FPS for the minimum possible latency, but naturally it can only display at 60 FPS, 75 FPS, or however many frames are allowed by the display hardware. The user would see this framerate, not 24 FPS. However, due to the additional steps needed to get the Leap Motion data into the game (the camera motion capture, data processing, etc.), there is a delay on top of the regular latency caused by the game FPS.
Finally, since the game thread is running at 120 FPS, it is able to run concurrently with the tracking data. This doesn't introduce any additional latency on top of the 120 FPS already accounted for by the Leap Motion camera.
As a result, the delay that you experience from hand motion to photon would be <35 milliseconds*, but the game itself should run just fine. This 35 ms figure is one that we're working to bring down, both through optimizations on the software side, and in the long term through new hardware releases. Regardless, even with 35 ms, we've generally found that users have a solid experience, since the requirements for hand tracking are somewhat less stringent than those for head tracking (which has to be 20 ms or less to avoid nausea).
( * Just noticed a typo in that section of the VR Best Practices, as it should be less than, not greater than! Will fix.)
