Both NodesEnum::nodes() and protobuf::ParseResult::nodes() don't have documentation, although the latter hints
// Note: this doesn't iterate over every possible node type, since we only care about a subset of nodes.
Is the goal just to provide a helper function to extract ParseResult::{warnings, tables, aliases, cte_names, functions}?
I'm trying to do some more detailed parsing of a statement tree, and would like a "visitor" pattern. I need more information than scan() provides. Is it useful to extend the nodes function?