It's early days yet but this is exactly what I'm trying to achieve with stash[0], plus the ability to browse notebooks over HTTP and run/embed dynamic queries into documents. It's essentially a specification of a directory structure for plain-text files and a bunch of tools for doing bulk manipulation. I began migrating all of my notes from Evernote a few months ago and it seems to be handling it well.
It uses a small-step interpreter [1]. Rather than capturing the interpreter's state on the native call stack, small-step interpreters explicitly model said state using stacks of objects - one per active AST node - and environments. Because they're no longer bound to the stack of the implementation language, execution may be cycled, paused and resumed at will.
Thanks for the feedback, I like this idea. I can imagine it working well in 256-colour mode, gradually increasing colour intensity in line with write-frequency.
[0] https://github.com/no-silo/stash-design, https://github.com/no-silo/stash-server