I think there are a lot of theories as to how to make temporal table systems work. It hugely depends on your requirements. That said, the data warehouse community has built a kind of solution with dimension tables representing time, and fact tables aligning to those dimensions via relations. It makes certain temporal "grain size" problems much easier to solve (and usually faster). Though using Pg's date extraction functions I've gotten pretty amazing performance as well: basically creating on-demand time dimensions as needed.. Read Ralph Kimball's work on data warehousing for a good introduction.
This isn't the same as creating versioned tables per your references above but I hope will be useful in your research..
