the difference between the spring example you have and what we are zeroing in on is the extra energy that can be had durring the process. ie.... the HHO
we are paying for the energy we use as we go , and generating extra energy durring the process.
It is the same principle.
In my mechanical equivalent you insert energy to extend the spring. The contracting spring can be used at any time to get SOME of this energy back.
That's the same as your idea of using energy for electrolysis and regaining this energy at a later time.
In both cases the problem persists: reducing the density of anything (Molecules, inflatable balloons, mechanical devices) in some medium will need exactly as much energy as you are gaining by letting it lift by buoyancy and fall down again.
My example uses just a mechanical device to see more clearly why it takes so much energy.
Just keep in mind: you will need more energy for splitting water to HHO then you can regain by recombination of HHO to water.
(At least as long as the splitting is done under higher surrounding pressure then the recombination).
Couldn't this be dissolved gases, which are released in a low pressure environment? Sounds to me like the stuff happening to divers when they inhale gases at high pressure and then surface to lower pressure.