It would be kind of silly if you couldn't group up devices like that. If you think about it, modern networks work very much like this. Take a modern-day setup. We have companies with servers that all slave a number of computers to themselves. Those computers, in turn, will slave other devices (such as flash drives, mice, keyboards, monitors, ect.)
Your rational is correct. You can, indeed, create groups of devices, slave them to a commlink, and then have the decker slave the commlinks to himself. This creates a mobile PAN (personal area network) mesh that the decker can more easily police and protect, since he's the first thing an enemy hacker has to get through. There are, of course, problems with this setup.
The biggest flaw to this setup is what happens when someone DOES make it through your detection and firewall unchecked? The answer? They now have EVERY piece of hardware your entire team owns at their fingertips. And until you can get them out of your node, they can choose to screw with any of the devices. Normally, if they want to hack the street sam, they'd need to be within range and hack their commlink separately. It's easier, but that's multiple points of failure, since the enemy hacker can't get to anyone else on your team through the sam. With you protecting the group, you're more likely to catch any intrusion, but if you fail, the enemy hacker now has three marks on your entire team and hilarity can ensue (read: bad things for your team).
Generally, you want to protect all of your party members and slave all of their devices to your deck. Period. There is no reason NOT to do this, honestly. The chances that you'll miss a hacker in your system are pretty low anyways, and even if you do miss them, rebooting your deck will sever the connection and erase the hacker's marks on your deck, forcing the enemy hacker to have to start again on someone else's device.
No.
A small number of new options are added (~20 pages worth) and a lot of super hazy talking is done about the matrix with lots of 'kewl' formatting and fiction inserts, none of which is above the level of 'getting together with your gaming friends and kibbitzing over a few beers'.
It is not a product which clarifies the matrix rules in any manner. If you want to run things with matrix stuff in sr5, my advice is to simply create a houserules document answering those questions for your group. It's sub-par, but it's what you will get for the foreseeable future.
I very much doubt any future products made by Catalyst will have clearer matrix rules or a clearer view of what the matrix is.
Additionally, you'll find that for anything but the most min/maxed characters, the dice pools and amount of rolling will result in dead deckers or defeated hacks far more often than not. Rewriting portions of the matrix chapter to more accurately reflect the hacking dice pools you expect in your game will likely solve many of the difficulties you are having.
Best Answer
On page 246 of the core, under the Agents heading, it says
Therefore even though it counts as its own entity it is still a program of a specific type. You can however have multiple different agent programs installed, but they cannot run simultaneously and requires a system reboot to switch since it counts as a persona. The section then goes on to state:
So to continue to explore that, the Pilot Programs section states that:
Therefore I would say that teamwork is possible with an agent, but you do so at a risk. If the target action has a high difficulty rating, the agent program might have to make a test just to attempt to help. With the right specific commands it can be prepared to help, but I imagine that would take some preset conditions to let it know when to pile on without trouble.