Best strategy for incremental development of a very large world and what are the limiting factors of large world development?

That’s been the issue since always. Nearly all other “problems” are BS made up by people that don’t want to work.

Start by defining what system you want to support.
Make a random 4 tile landscape with an auto material. (Use a decent size tile, 4km squared in 2x2 configuration).
Bench performance on that system.

If it works (meaning 120fps at the desired size) - which it probably won’t because the engine sucks at performance, then you can move forward with that tile size.

If it doesnt, you have to pick out a different tile size.

Or you can go forward knowing that at the end of it all youll need to convert from using the landscape to using meshes.

After that, you just take the process one tile/level at a time.
On avarage, a 4km tile takes about 3 days to dress up so it isnt empty…

Obviously, your numbers are wrong but they also havent mattered one bit since always.
You just enable world origin rebasing or create your own rebasing system and the overall size can be infinite.