Thank you for the ideas!
Now I finally got to this thread and see that it has at its core the same idea as I had for Linear Cross-Contract Storage (LCCS) for Ethereum 1x State Rent proposal. I identified it as not a “nice to have”, but as a necessary primitive for most existing contracts to transition into the State rent regime. I see that LCCS makes Actor-Asset implementation almost trivial. One difference that I see is this: In LCCS proposal, I specified that the cells are only readable by two entities: by owner, and by writer (actor in your terminology). The reasoning behind this is to allow contracts (mostly actors) to collect CALLFEEs for calling them, so that they can pay rent for their code and account info. If it is allowed for anybody to read the cells, then it would be possible to circumvent this and deprive the contract from this revenue opportunity, and therefore, from the ability to pay rent for its code. I will expand on this in the next version of the document.