Markdown formatter refactor

[milmazz]

I was curious about this new formatter in ex_doc and I did an initial review. Please let me know what you think.

[github-actions]

📦 Docs artifacts are ready: https://github.com/elixir-lang/ex_doc/actions/runs/21784421513/artifacts/5418153935

[josevalim]

Hi @milmazz! There are some other PRs and I have also refactored some of this code locally, so let's please hold on before merging this, unless there is a bug fix. :)

[milmazz]

Hi @milmazz! There are some other PRs and I have also refactored some of this code locally, so let's please hold on before merging this, unless there is a bug fix. :)

No worries, I'm so inactive in this project that I wasn't planning on merging code without any maintainer review.

[josevalim]

@milmazz I am done with my refactor. Please rebase! The markdown module changed considerably but maybe I dropped a few things :)

[milmazz]

@josevalim Hey, the new formatter is basically a whole revamp on the previous version :)

I just did some minor refactoring:

• The optional description in the llms.txt should be a block quote according to the docs from: https://llmstxt.org
• Group the resources just once and pass that as argument to private helpers
• Use some pipelines to denote the sequence of data transformation
• Use File.write!/2 for everything, as in the html formatter.

[josevalim]

I decided explicitly to not go down this route so we can more clearly see what is passed to what and what depends on what.

[josevalim]

Should we make config the first argument for consistency?

[josevalim]

One tiny comment and we can fix it!