When trying to see why headers are not showing once I created some folders and converted some groups/files to folders, I started exploring the .scriv/ directory in BBEdit. Looks like text contents of the folders themselves seem to sit in BinderStrings.xml, while nodes are in numerically named directories. Is there such a difference in where the text is stored depending on the type of the node?

The stuff in the BinderStrings is actually Scrivener’s search index data storage area. All of that is raw plain-text, unformatted, and is used to keep project searches snappy. The actual formatted data is stored in the serialised RTFD files. It doesn’t have anything to do with node type. Folders are assigned serial numbers just like files, and in fact operate identically as files do (which is how you can convert between the two so easily). Both have RTFD files, including whatever other auxiliary files they require, such as synopsis text files or notes.