Support for multiple, interacting swarms in the same environment

[mountaindust]

This requires two areas of focus:

1. Simultaneous updating up multiple swarms when an environment method is called.
2. Plotting multiple swarms in different colors on the same plot.

Because of the second part, this should probably not be done until #40 is addressed.

[mountaindust]

Revisited this while reconciling the documentation against the code for 1.0.1 (checked on master at 3792416). Bullet 1 looks done; bullet 2 is not.

1. Simultaneous updating — appears complete. Environment.move_swarms(dt) moves every Swarm in envir.swarms and then advances the environment clock once. Verified empirically with two Swarms (sizes 20 and 30) sharing a Brinkman-flow Environment over 5 steps:

Both Swarms advance.
Each keeps its own pos_history, with length matching envir.time_history.
envir.time increments once per move_swarms call, not once per Swarm.
Swarms with different apply_agent_model subclasses coexist without trouble.
Calling Swarm.move() directly on one Swarm emits the appropriate guard: "Other Swarms in the Environment were not moved during this environmental timestep."
2. Plotting multiple Swarms on one plot — still open. Swarm.plot and Swarm.plot_all never reference self.envir.swarms, so each Swarm renders on its own axes. This is the piece noted as waiting on #40.

Interaction — still open, and arguably a third item. There is no built-in mechanism for agents in one Swarm to sense or respond to agents in another. A subclass can reach the other Swarms through self.envir.swarms inside apply_agent_model and implement it manually, but nothing in the library supports it directly. Since the issue title says "interacting", this may be worth tracking separately from the plotting work.

Related doc fix: docs/quickstart.rst previously stated that multiple Swarms in one Environment "should be considered broken for the time being." That was inaccurate. It has been corrected for 1.0.1 to describe the working move_swarms path along with the two gaps above.

Suggest narrowing this issue to (a) joint plotting and (b) cross-Swarm interaction, and closing out the simultaneous-updating bullet.