Can you post some before and afters of your editor showcasing the feature?

How the highlighting works without the feature:

Here's how it looks like with the semantic token support:

We can see in the second screenshot that the "document" parameter is highlighted differently(the :Inspect from neovim is on "document" parameter). Next-ls has sent additional information to the editor for this token @lsp.type.parameter.elixir which clients can override and display in their own way. (ignore the coloring since it's the default from my theme and looks the same as atoms).
The goal of this PR should be to identify other "special" types of tokens that would be useful for developers to customize before adding it as a feature.

I'll leave two articles on the semantic highlighting from VSCode:
https://code.visualstudio.com/api/language-extensions/semantic-highlight-guide#:~:text=Semantic%20tokenization%20allows%20language%20servers,the%20syntax%20highlighting%20from%20grammars

And jetbrains:
https://blog.jetbrains.com/pycharm/2017/01/make-sense-of-your-variables-at-a-glance-with-semantic-highlighting/

A nice usage could be adding a different type to macros such as @lsp.type.macro.elixir. Let's say for an example we have:
Logger.warn "a warning", this type would be added to the warn token allowing the user to customize how the colorscheme works.

Note that we would want to be careful here since all operators could easily be interpreted as macros as well breaking all colorschemes in the process. We would probably want to exclude any Kernel level macros from this.