What exactly is being tested here?? The amount of applications that can be launched and suspended (typically limited by RAM) or the number of applications that can actually RUN simultaneously (typically limited by CPU power and in a small way RAM).

Most applications are sitting waiting some sort of input so I would class them in test one - yes they are not suspended by the OS but since they are waiting for info they are suspended by themselves.

The best way to test scenario one is to run an application that actually does something all the time, then run another instance, the another, then another and so on.

I remember in the days of windows NT and windows 95 there was a simple bezier drawing test exe that microsoft used to flaunt, that showed you could run multiple instances of it. At the point where CPU bandwidth became an issue you would see all of them slowing down. We really need an application like this to show the performance of multitasking between the phones.