As semgrep is used by more and more users we need to have a way to develop rules for new languages/frameworks as experimental support without breaking the main rules that everyone else uses.