Worst case a few sats get lost or the booster doesn't land correctly. As long as they test this on their own sats first (which are rather cheap) I do not see the issue.
They will want to test to breaking point after all.
So yeah, seems that they want to reduce their starlink cost
No people were hurt and no windows were broken.
Even managing propellant tanks is complicated dilema. Take a big jug of water that is half full and slosh it around. Its one of thousands small things that needs solving. And solutions need to be implemented with as few grams as possible.
I think the SpaceX organization has proven to be prudent risk takers. They have had many spectacular failures, but they have been good at communicating the risks ahead of time.
Just starting a rocket engine is a complicated dance that can easily turn out badly if not handled with extreme precision. Sorry, but I don't think there is any way one can equate the complexity of a diving bell with that of a rocket.
A rocket has to deal with rocket thrust/structural cross-section units of pressure. I'm neither a engineer nor informed about the structure of a Falcon 9, but that number is > 1 atm even if the entire cross-section is structural.