Add support for "distributed attacks"

[whotwagner]

It should be possible to run multiple attackm8 agents, that connect to one central instance. That instance runs the commands and must be able to select on which instance the command should be executed.