it is not yet, but getting very close to accurately being regenerated again. pottedmeat has done virtually all of the work on the parser to make it pickup every little piece of doc, and has almost finished tweaking the API tool to mix and munge it all together (a very difficult though important step in disambiguating all of the weird code styles that exist in core/base, and all the inheritance going on in dijit) ... once the API tool is ready, we should be able to (in theory) regenerate that data once again.