Trailing comma in AST

For background, Stripe uses rubyfmt to format Ruby. One thing rubyfmt doesn't have today is trailing commas: all trailing commas are stripped by formatting, which is the preference of the maintainers of that project.

For a while I've wished for something like "rubyfmt + trailing commas." Before it even makes sense to have the conversation about whether to make that change, I was thinking through what it would even look like in practice. One thing which would make things easier to adopt would be if rubyfmt could run in a mode that preserves trailing commas, rather than universally adding or removing them (you could migrate slices of a codebase at a time, instead of having to land a PR touching hundreds of thousands of files all at once).

rubyfmt now uses prism for parsing Ruby files, and prism omits information about whether various constructs had trailing commas.

I was curious to get a temperature check on whether this is something that prism would ever want to see tracked. For example, in ArrayNode or in HashNode, to have some way to know that the user had written a trailing comma.

This is a very "what if" sort of line of thought, so please don't feel rushed to answer, and also I completely understand if the answer is definitively "that's not something prism is open to incorporating."

Start by reading Prism's ArrayNode and HashNode handling and how rubyfmt consumes Prism's AST. Investigate whether trailing-comma presence can be represented for these nodes without changing existing formatting behavior. Done means documenting a concrete recommendation on whether Prism should track this information.