"Type" is probably not the best term/description... The last type will be None if the trace of the variables failed. Another feature is all of the data is stored in a list of tuples called log.
I'd still like to add a couple of more features. Ashutosh Mehra mentioned some issues around the use of EncodePointer. Simple scenarios are not that hard but anytime more functions or APIs calls are added to flow tracing logic things get complicated quickly. Also, it would be cool to solve this problem. I tried some approaches of adding sections and patching the IDB but I was unsuccessful. If you have an ideas, comments or find bugs please send me an email or ping me on Twitter.
Most of window/viewer code came from the post Extending IDA with Custom Viewers. There are a lot of great post on the MindShaRE blog. I'd highly recommend reading through them if you haven't already.