Before I review the changes themselves can you elaborate on the technical need for this change?

To me it seems like a subjective change in UX. I think that having the ability to set data at compile time is important and useful. At the same time being able to delay that via a constructor may be useful as well.

Without deeper consideration at this stage I'll await your response and instead superficially propose the idea of having 2 ownership standards. One for configuration and one for construction. That has trade-offs as well but I don't think that's as much of an issue as providing one UX over the other.

This is something I considered, but there are a few caveats to this.

1. In the future we expect to have the ability to import storage from libraries, thus enabling the ability to set the owner at compile time
2. With most developers coming from Solidity, a constructor is something they are already accustomed to
3. If a developer chooses to set the owner at compile time, they can do so using just the SRC-5 standard. This library is intended for convenience and is by no means the only way to implement the standard

I believe the proposed changes make Sway code significantly more readable and less verbose. Having to call storage.owner.my_function() for absolutely everything simply doesn't make sense compared to just the function itself. The trade off being well worth the minor inconvenience of a constructor in turn for easier auditability.