Based on my experiences...
Based on those I know, the guidelines you link to are not general.
The day-to-day supervisor approach is typically implemented passively because the main supervisor (i.e., the guy in absentia most of the time) is too busy to actually manage the students. For this approach to be practical, the day-to-day supervisor must have expertise in your research and must be aware of the general research path of the group (i.e., can provide not only advice but direction).
For what it's worth, most labs with a day-to-day supervisor in addition to the main one will be more difficult to work in as a graduate student. The only case where this can be a benefit is where the formal advisor is some ridiculously well-known figure in the field, and your just being in their lab will lead to opportunities down the road. For the majority of labs with this setup, though, it's simply because the formal advisor is too busy to deal with (or otherwise disintereted) the graduate students, and has set up someone else to deal with them. This typically leads to communication issues, lack of guidance, long delays in your advisor reviewing your work & publications, and frustration. I would approach these setups cautiously.
Edit based on comments: The main reason these setups tend to fail is that the supervisor's unspoken job description is one that will never be filled; replicate the domain expertise and research experience of the advisor while essentially being a graduate student counselor. Anyone who can do that will be running their own lab, not helping you manage your students.
This means that those who do take the job either don't have the relevant domain expertise to adequately answer student questions, or relevant research experience to design, run, and analyze data from a complete research project. Any lacking expertise translates into "lets just wait until your next advisor meeting", which adds long delays to everything.