Future of elixir_sense integration

I get the impression from my reading of the code (I've done quite a bit of digging around now) that elixir_sense is on the way out.

I wanted to see if I could make completion a more acceptable speed, and so I made a bunch of changes in elixir_sense to reduce redundant work, add caching for module and function lookups (invalidated based on file changes) and made a few smaller changes through the completion processing pipeline outside of elixir_sense too.

I got performance on my slow laptop from 300-600ms for a completion down to as low as 90 and generally around 100-150ms.

However I noticed there seems to be quite a lot of duplication of the things elixir_sense is doing, ast parsing, document handling etc. And with an overall much nicer design. Elixir_sense is still doing a lot for this work, despite it mostly already existing in the elixir_sense codebase.

Would I be correct in thinking that elixir_sense is being partially or full fazed out? Or at least altered to operate on the Forge ast and documents?

If so I'll focus on adding a module cache to the Env and passing that over to elixir_sense so it's easier to migrate away from it in future. I'll also see what I can do as far as avoiding redoing parsing work on the ES side.