Performance of Agent-Chains

This update includes a couple of bugfixes and a performance update for the advanced flocking algorithm.
(Also as a small preview: The next update will be of a different kind. I will create a version, which shows the actual performance of the different flocking algorithms!)

Advanced Flocking - Follower Leader Chains:
Two updates ago I told you about how the navigation for the advanced flock works: All the agents are following the leader, who uses pathfinding to find the endpoint. But when the agents lose sight of their leader, they try to follow it indirectly, by following other agents, that still can see the leader. Building so called follower-leader-chains, the agents can follow their leader indirectly over multiple other agents.
But as it turned out, my process of calculating those follower-leader-chains was super slow. So wrote a new algorithm that calculates the chains much more efficiently and even correctly! The result is: When there are long chains with many agents, the FPS are tripled!

Bugfixes:
- Agents know their own collider again
- The single behaviours of the flocking algorithm are re-weighted
- While changing the obstacle height in the map creation, the correct textbox shows the height-value.

(02.11-12.11)