There are two technologies related to sleeper agents: one, identifies which systems are the sleeper agents on and the other... I'm not sure what it does... does it eliminate all the sleeper agents at once per elimination program activated on systems? that is, I still need to activate the elimination program one by one decreasing happiness on each single system, right? 

There are 3 actions (2 of them are researchable, but mutually exclusive). Basic reverts one sleeper after 5 turns, if present, but system happiness is lowered by 25 for this period. Next one reveals all sleepers present, it is active for 5 turns and costs just fixed amount of influence. Basically, it allows not to hurt your system with base revert action, if no sleepers are present. The last one just kills off all sleepers present at once, also costs just influence.