Of course, you are absolutely right that your approach is faster, but, as you point out, it reauires complete control of the entire process. Also (which I did not make clear enough in the article, so please excuse me), the appraoch that I am suggesting is very much a development technique. Our data warehouse is evolving constantly, and dimensions and attributes are being altered all the time (this is one of those projects where "Agile" means less design and planning than is healthy...) so the design is not stable enough to persist dimesnion attributes - yet.
Once we reach a more stable setup, then we will certainly adapt the process to persist unchanging dimension and fact data, and adopt the most appropriate techniques which suit the project.
Thanks for the very succinct and clear description of how to go about doing this!