Cleanup and restructure EpsteinCivilViolence and PdGrid examples

[EwoutH]

Cleanup and restructure the EpsteinCivilViolence and PdGrid advanced example models. This PR is similar to what was done in the basic examples in #2365.

It draws the discrete spaces as was supported in #2386 and updates the imports to be absolute, similar to #2402.

EpsteinCivilViolence and PdGrid now:

• Have all their Agent code in an agents.py file.
• Use a flat structure.
• Use the newest SolaraViz API
• Have the old visualization code removed

A few things to note:

• Handeling agent_portrayals is still finicky
• I couldn't get the square shape to work
• EpsteinCivilViolence shows remarkable stable behavior. Is this correct?
• Claude 3.5 Sonnet went full YOLO and found a way to get a choice selector to work:

  "activation_order": {
      "type": "Select",
      "value": "Random",
      "values": PdGrid.activation_regimes,
      "label": "Activation Regime"
  }

  @Corvince Is this a proper approach? If so, does that mean that Viz: Implement Choice UserParam #2376 isn't needed anymore?
• Real interesting changes in behavior between the random, sequential and staged activation regimes! Sequential and staged quickly go to all cooperating or all non-cooperating. Maybe we should double check if this works as intended.

[github-actions]

Performance benchmarks:

Model	Size	Init time [95% CI]	Run time [95% CI]
BoltzmannWealth	small	🔵 -1.6% [-1.9%, -1.3%]	🔵 -0.1% [-0.3%, +0.1%]
BoltzmannWealth	large	🔵 -2.0% [-2.4%, -1.6%]	🔵 +0.0% [-0.3%, +0.4%]
Schelling	small	🟢 -3.9% [-4.3%, -3.6%]	🔵 -0.1% [-0.4%, +0.2%]
Schelling	large	🔵 -3.7% [-4.8%, -2.6%]	🔵 -1.9% [-2.9%, -1.1%]
WolfSheep	small	🔵 -1.8% [-2.1%, -1.5%]	🔵 +2.7% [+1.6%, +3.7%]
WolfSheep	large	🔵 -1.7% [-2.1%, -1.4%]	🔵 +3.2% [+1.8%, +4.6%]
BoidFlockers	small	🔵 -1.1% [-1.5%, -0.6%]	🔵 -1.9% [-2.7%, -1.1%]
BoidFlockers	large	🔵 -1.0% [-1.6%, -0.3%]	🔵 -1.7% [-2.1%, -1.2%]

[quaquel]

The code reorganization etc. looks fine from my end, so I'll approve the PR on that ground.

EpsteinCivilViolence shows remarkable stable behavior. Is this correct?

Luckily for you, I wasted an entire morning on this for my teaching version of this model on this question: swap out OrthogonalMooreGrid for OrthogonalVonNeumann grid. There are a few more things I would like to do with this example (but those can wait for a separate PR).

Real interesting changes in behavior between the random, sequential and staged activation regimes! Sequential and staged quickly go to all cooperating or all non-cooperating. Maybe we should double check if this works as intended.

I plan to do a pr on this soon based on an mesa 3.0 updated version of my teaching version. It might be nice to include the activation order as a user input in the gui version. It would not be strange if activation order matters a lot in this model.

[EwoutH]

Thanks for the swift review!

Luckily for you, I wasted an entire morning on this for my teaching version of this model on this question: swap out OrthogonalMooreGrid for OrthogonalVonNeumann grid.

Huh, that's indeed interesting. Tell me how that went tomorrow!

It would not be strange if activation order matters a lot in this model.

Yeah definitely play with it (or I can show it tomorrow). Kind of strange! (but also haven't gave it proper brain cycles, those are scares this week)

---

I will leave this PR open for a bit for others to take a look at it, we can merge tomorrow.

[EwoutH]

@quaquel merged, go ahead and follow up with changing the OrthogonalMooreGrid to OrthogonalVonNeumann. If you figured out why that mattered, would be very curious to hear that!

[EwoutH]

@quaquel so we don't forget it, should I create an issue for changing the OrthogonalMooreGrid to OrthogonalVonNeumann?

[quaquel]

Yes