I didn't read this post earlier because picking idle problems from general problems of running are hard enough.
I get a totally different opinion of this case. I almost suspect a single cylinder misfire, usually caused by plug wire ends. The key is always to isolate the problem. In the other post the key was two intersecting symptoms. In the shop, if I didn't have experience as happened the first time I found that problem, I would see if there was a single cylinder misfire first.
All of the CIS systems have a couple unique features that can assist in the engine diagnostics with no tools at all (well you do need a finger and a twelve millimeter open end wrench.
The first thing I do is remove the airfilter and gently push down on the airflap while the engine is running. If this totally smooths out the engine there are a couple possibilities. First is there could be a general mixture problem, second the injectors could be variously restricted; a motor running in closed loop (figure out what that means because it is the heart of modern control) will be running so close to lean misfire that any difference in injected quantity will cause such a misfire in the weaker cylinders.
Unfortunately. it might also lead one to some inproper decisions as in the other car with a single cylinder lean misfire caused by vacuum leaks to one cylinder. When the common mixture is raised enough to get that one cylinders mix in range it will also start running again.
The other feature of CIS is the ability to do cylinder balance testing by openning the injection line similar to a diesel. By cracking the lines loose at the fuel distributor one at a time, the relative power of each cylinder can be evaluated. Cylinders that don't cause further roughness when opened are not running strongly.
A little sensitivity here and some follow up electronics (see the DIY article; evaluating engine controls) can get to the bottom of this.
In the vacuum line case, the above testing would have found the cylinder that got the leak (#8 if I recall). The further testing with the airflow plate would probably made the cylinder run again showing it to be a mixture problem. It then might have been tricky to decide that the mixture problem was vacuum leak related. Not too difficult though, because the problem goes away at higher speeds. This kind of mixture problem is almost always a vacuum leak because the amount of leak stays constant and as the engine speed increases it becomes proportionally smaller till of no consequense. Single cylinder mixture problem caused by fuel distributors or injector flow restrictions don't go away at greater flow rates, often they get worse.
Bosch Master, ASE Master, L1
33 years MB technician