overall looks good, but most of the code is duplication from the original versions. I wonder if we could refactor to share the logic?

I tried refactoring out the common structs and code logic to form a "clean" solution (the latest WIP commit). However, with this, we break the ABI as the accounts are now passed in differently for the original initialize and release_inbound_mint instructions.

If I were to preserve the ABI, there is minimal/no refactoring possible for the structs. This makes it difficult to use refactored code logic as there is no common shared struct that can be used; instead every required account has to be passed explicitly. Although I prefer the "clean" solution as it reduces code duplication and makes the -multisig variants/ any future variants easy to follow, its downside of breaking ABI makes it not super viable.

Ideally, we'd want to refactor code without breaking the ABI.