I want to create an incident on measure "CPU total time" that will apply on selected hosts,
but i can't filter the measure by host.
The measure is under-
System Monitoring -> Host Performance -> CPU Total Time
I tried using "CPU Total Time" under "Process Performance" but it doesn't returns the same values.
One limitation with the host level measures you'll find within the system profiles is that they are essentially across the board with no way to limit them to specific agent groups or agents. So if you define an incident it will have to apply to all agents that are in that profile - the extended email action plugin is an option to experiment with filters.
What I would recommend though is to look at the alerts that are available at the global infrastructure level. This allows you to use labels and host groups so that you can define host health alerts with much more detail as to what entities should be considered:
I'd take a look at that.