We have duplicate tree-sitter.json files registering the different grammars and their supplemental files. This information is expressed in the project root and again in each parser. We need to see if tree-sitter can consolidate down to a single file now, and possibly open issues to get some improvements where we get stuck.
Much of this is because we are a single git-repo that hold several related grammars and we package and deliver those grammars as collective bundles. This isn't often done and causes friction from the expected project setup. Just because we find bugs doesn't mean the tree-sitter project is going to make changes to support us, but it would go towards documenting why we are setup the way we are.