Which approach to rust is more idiomatic (Helix vs Zed)?

Hi! I am curious what is the current "meta" (by "meta" I mean the current rust's community idiomatic and best-practice way of doing things) of rust programming. I am studying source code of 2 editors I am using: Helix and Zed. And I can see that while they are doing a lot of similar things (like using LSP and parsing it outputs for example) the code is kinda different.

It starts from the file structure: in Helix there are not that many folders to look at (like you have helix-core which contains features like "diagnostic", "diff", "history", etc but in Zed every single one of them is a different crate , which approach is more "idiomatic"? To divide every feature as a separate crate or to use more "packed" crates like "core".

Then the code itself is kinda different, for example I am currently looking at LSP implementation in both of them and in Helix's case I can follow along and understand the code much more easily (here is the file I am referring to. But in Zed's case it is kinda hard to understand the code because of "type level programming" stuff like this one for example. It also doesn't help that files have a lot of SLOC in them (over 1500 in normal in Zed's repository, is it also how you do rust?) Maybe I am just used to lean functions from other languages (I mainly did TypeScript and Elixir in my career).

Other thing I see is that Helix has more comments about "why the thing is doing that in the first place" which I find very helpful (on the other hand in seems that Zed's is abusing a lot of "type level" programming to have a self-documented code but it is harder to reason about at least for me) which approach here you prefer?