Swarm experiments

[lmas]

Refactor

• Ensure swarm is completely decoupled (still uses a global rand for example rand is only used during init for spawning boids, so should be fine)
• Might finally have to clean up the boid constants and make them swarm settings (so individual swarms can have custom behaviours)
• Enforce max amount of neighbour boids a single boid will match with (and run expensive distance rules against)?

Experiments

• Benchmark baseline
• Investigate performance when running multiple swarms
• Might want a "leader flag" to enforce boids following single leader and not be split up into multiple groups in the same swarm
• Investigate having half the boids act normal (leaders, runs expensive boids rules), while other half only tries to catch up/match with leaders (followers, less expensive rules?)

Source suggestions