There are so many issues with either a pod or a solid tube. The biggest one is buoyancy. If it is full of air it will need to be weighted, and have air added/removed to adjust for pressure changes with depth. But imagine if the kid "turns turtle". Any brief head-down orientation will result in the fabric collapsing around the kid's head as the air rushes up to his feet. Correcting this requires forcefully turning the kid around to keep his head up. This is one (of many) reasons why commercial divers in dry suits use solid helmets. It isn't an issue for wetsuits. I'd rather slap a full helmet on the kid and tie him up into a bundle. It won't be pretty but does solve the buoyancy issues. I'm sure they are also considering drugging the kids to reduce the risk of panic attacks.
Here is the nasa device, the personal rescue enclosure:
Of the difficulties the pod solution poses, I don't think that neutral buoyancy us one of them, so long as the pod is rigid (e.g. steel) in construction.
The best succinct definition for neutral buoyancy I could find was this one, from Wikipedia:
> ...a condition in which a physical body's average density is equal to the density of the fluid in which it is immersed.[0]
This means that once the pod has been weighted to be neutrally buoyant it will be so at any depth of water, since neither it or water are compressible[1] and thus their average densities are unchanged by depth.
The challenges of maintaining neutral buoyancy arise when you submerge a compressible object, such as a scuba diver wearing an inflatable buoyancy control device (BCD). The water pressure (caused by the weight of the water above) will squeeze them and their equipment, compressing anything that is air filled.
To sink from the surface, the diver has to deflate their BCD, and then put sufficient air in it to achieve neutral buoyancy at their desired depth. If they swim downwards, the higher water pressure will squeeze them some more, the air in their BCD will compress some more, and their average density will increase. They will now be negatively buoyant and will sink.
The deeper they sink, the higher the water pressure will be, the more the air in their BCD will compress and the more negatively buoyant they will become. So they'll sink faster and faster. This happened to me once when I was first learning to scuba dive. It can be scary.
The opposite applies to ascending. This is why good buoyancy control is such an important (and difficult) thing for scuba divers to learn.[2]
From a buoyancy perspective, a metal pod is more like a camera in an underwater housing. These are rigid in construction (usually either plastic or metal) and their buoyancy characteristics don't change with depth.
None of this is relevant because the cave contains a restriction so tight that a rigid pod wouldn't fit through. The last thing they want is a solid object wedged into the restriction.
The SpaceX design is apparently flexible to fit through the cave, so some buoyancy adjustments would be necessary.
With a flexible pod, you'd partially deflate it to get it through awkward areas, which would make it negatively buoyant. Once re-inflated to the previous pressure it would be neutrally buoyant once again.
Re-inflation could be done via the same gas cylinder that's being used to replenish the air in the pod.
Here is the nasa device, the personal rescue enclosure:
http://www.astronautix.com/r/rescueball.html
(It wouldn't work. Wrong shape and no buoyancy control.)