Reversibility of compressed air sinking in water

This state of affairs can be analyzed without including entropy arguments, as follows.

Submarines deal with this problem all the time, in that if the sub is positively buoyant and is set to descend deeper under power, as it descends the pressure of the water outside the hull forces the hull to contract, decreasing its displacement and hence its buoyancy. So the deeper it get, the less buoyancy it possesses- and if it dives beyond the neutral buoyancy point, it will sink on its own even with its motors stopped. If unchecked, the sub continues down out of control until its hull gets crushed and implodes.

Alternatively, if a dived sub is trimmed into a neutral bouyancy state and then turns on its motors to drive to the surface, as it rises its hull expands, increasing its displacement and hence its buoyancy- leading to an ever-increasing rate of rise which persists even with its motors shut down. If unchecked, the sub rises faster and faster and will breach itself partway out of the water upon reaching the surface.

This means that a dived submarine is dynamically unstable. If trimmed into perfectly neutral buoyancy, a slight upwards perturbation will grow into a rise, and a slight downward perturbation will devolve into a dive.

A dived sub must therefore be driven "hands-on" at all times, with both its net buoyancy adjusted on its way down and again on its way up, and the amount of lift its diving planes generate being similarly trimmed constantly by changing their angle of attack and the sub's speed through the water.

Not my strong suit, but...

If the container is compressed vertically instead of horizontally, the hydrostatic pressure would be reduced allowing the vessel to rise again. Once it crossed the threshold required for the compressed air to return its buoyancy, even if the squeeze was released it should continue to rise as the gas expands.