[ufmace]

I think it's a few things:

They're already highly confident that if they have sufficient control over the booster trajectory they can execute the chopstick catch, so they don't particularly need to demonstrate that part more. Executing a pseudo-landing at sea lets them validate their booster flight controls perfectly fine without risking the launch tower and associated hardware. They can also do stuff like stretch the trajectory and control mechanisms to their limits to see how much they can handle, and not too big a deal if something goes wrong. Presumably any actual landings will be well within the known safe limits on all parameters.

I bet this first booster also has a lot of minor weird things associated with shaking out the manufacturing process, and they don't entirely mind testing to destruction the first one to get rid of it permanently and use the ones coming from a more proven manufacturing process for important work.