Improve Composite Agent state-machine for improved integration of mobility

[andreiolaru-ro]

We will have the following states:

• (zero) -- there is nothing; the agent can be checked & pre-loaded;
• -- new -- the agent is constructed
• STOPPED -- the agent is created (the constructor has returned)
  • components can be added
  • the agent can be transferred
  • the agent may or may not be loaded on a platform
  • the agent may have all intended components or not
• if parallel processes happen during which components should not be added or other processes may not happen, the agent enters the TRANSIENT state (most likely when it is moving)
• -- the agent is started (probably after being added to a platform)
• STARTING
• RUNNING
  • the agent is deployed, loaded on a platform, and prepared for application execution
  • the agent thread is running and processing events
  • simulation/execution can be started or stopped
• -- the agent is stopped
• STOPPING
• back to STOPPED

[andreiolaru-ro]

Ideally, components should be loaded at the time when they are added (after the agent has been loaded), and not when the agent is preloaded. This is however difficult as for agent there is a separate instance that loads them (the AgentLoader), but for components there isn't, so how should the AgentComponent.preload() method be called? Or the parentChangeNotifier()? Must think about it.

[andreiolaru-ro]

Completed in 67f4fac