using explicit language identifiers in generated markdown code-fences

[the-mikedavis]

Hello! 👋

Currently, markdown code-fences generated for documentation do not specify the language (example). I think this works fine in most editors since the editor will assume the code block's language is Elixir. Some editors don't have that behavior though and fallback to no highlighting. Other language servers I have used provide the language identifier explicitly: for example rust-analyzer and erlang_ls. It would be a very small change to use explicit

```elixir
```

blocks instead and I don't see any downsides.

What do you think? Would you accept a PR here and in elixir_sense that adds elixir to code-fences?