Very nice article. It provides a set of useful solutions to a common issue, while also highlighting the fact that some common "mantras" (i.e, Inline TVFs are more efficient than scalar UDFs) are not always true.

I'd have liked to see some charts showing how the different solutions scale from 1 to millions of rows, but that's really not something I feel should be required of a short article like this, and you've provided enough information for me to test this on my own if I want to take the time.

Finally one question: did you do your testing on a preview or RTM version of SQL Server 2012? If it's a preview version, there might be some debugging code slowing down the new function.