Re-scoping: this direction will proceed as a separate project
After reviewing how much of Myelin these designs would actually use, I want to correct the framing rather than let it stand.
Here is the honest accounting on 9th Oct:
The three options above need an EVM execution backend, an Ethereum state model and a new CKB-based sequencing protocol, none of which Myelin provides today. Direct code reuse would be limited to generic infrastructure: hashing primitives, a storage shell, the CKB follower adapter and some evidence-handling patterns. What genuinely transfers is the research trajectory.
The based EVM layer will therefore be treated as a separate protocol and project.
I have edited the title accordingly. Comments on where these assumptions are weakest remain very welcome.