EDIT: I reran the sagulator and found an error in my input (first sentence in italics is in error). The deflection is only 1/4". However, I believe the discussion of the dynamic nature of the loads and the behavior of the end points requiring a secure connection remain important considerations.
I just ran this through the Sagulator assuming 200lb total load and got a deflection of nearly 2". Statically, the wood can carry the load, but the problem is that your load is dynamic. Two or three kids could be jumping and shifting weight continuously. Your wood composite beam will probably not crack under their activities (you would need to use a waterproof type of glue to prevent its weakening as it gets soaked). The real problem is not vertical loads, but rather the fact that forces will come at many angles. As the beam sags under the weight it will have a tendency to bow and twist. Enough force in the right direction will cause the beam to rotate around its long access. This will result in the beam bouncing and rotating from the direct line across the pool perpendicular to the edges to something else. It is only a matter of time before it will shift one or both ends to the point it is not supported by the edge. There is great risk of someone getting hurt.
One solution to that is to properly secure the edges to the edge of the pool so that they cannot shift.
I would also consider that someone falling near the edge may land on the pool edge rather than the water. Cracked heads are no fun. I would make certain that at least the edges of the pool are properly padded and that the kids are wearing helmets.