Dimensional modelling was something invented by Ralph Kimball (although I'm sure Margy Ross had a great deal of input as well). They came up with the naming standards for different types of dimensions. On to be sure, there are also different types of fact tables (snapshot, transactional, aggregate/summary, etc), but I don't believe they ever referred to a "type-2 fact table". Interesting idea; although it basically seems like a variety of snapshot fact table, with different results over time.
Ralph Kimball has a nice intro to the three basic types of dimensions here: http://www.kimballgroup.com/2008/08/21/slowly-changing-dimensions/
Margy Ross has another article on the same site, which explores additional types of changing dimensions, usually by adding snow-flakes. As an aside, she downplays the numbering and instead names the different types.
A similar article introducing facts is here: http://www.kimballgroup.com/2008/11/05/fact-tables/
You can see that "type-2" doesn't really apply well to fact tables -- because there is no "type-1", etc. Instead, all you are really doing is creating what looks to me like a combination of transaction fact and snapshot fact. I'm not saying that's a bad idea -- in fact, I think I've done similar things myself in the past. Calling it a "type-2" though... well, you'll get all of the dimensional modellers popping on to tell you that it's a type-2 dimension, not fact. 😀
Oh, I ran into that same bug in SQL Server with merge output and relational integrity, and ended up solving it in much the same way too.