Hello again Tom, welcome... to the place where Rebol problems are solved...vs. exacerbated. 
Saw your Master's Thesis go by in the links-people-had-found. I was too tired to read it at the time, and put it in the pile of "things to read", but never got to it. Maybe now that you're here, you could summarize the major points or findings (in a way people who know the language already would best comprehend).
I don't know so much about stabilized. But the approach of methodically fighting each design flaw one at a time has proven fruitful.
We had a conference in 2019, so there's some stuff to watch which may be interesting:
Talks | Rebol [2019]
The big investment here is in the web build, which can run even things from way back:
Enter the Dungeon 💀 ...if You Dare!
But there's lots of other news. Something I think is a real success story is the combinator-based parsing engine ("UPARSE"):
Introducing The Hackable Usermode PARSE ("UPARSE")
So if you found historical Rebol PARSE to be appealing, you should love that. It's in the browser, too:
Visual PARSE on the Web... has arrived!
Building for WASI containers is a thing we can do, also:
Ren-C is Now on the Clouds ☁️ with WASI
Please feel free to post responses on anything you find here of interest. I try to go back and edit past threads to cohere things, or delete completely useless brainstorms (kind of like a wiki). On the downside this isn't a perfect record of conversation, but on the plus side some of the code in posts from years ago might actually work today.