fix(emperor-dilemma): replace relative imports and fix unseeded random calls

[abhinavk0220]

Summary

Fixes two bugs preventing the Emperor's Dilemma example from running, as reported in #337.

Bug 1 — Relative imports break Solara module resolution

app.py used from .model import EmperorModel and model.py used from .agents import EmperorAgent. Solara loads example modules in a flat namespace and cannot resolve relative imports. Fixed by switching to absolute imports (from emperor_dilemma.model import ...), consistent with how other examples in this repo are structured (e.g. bank_reserves).

Bug 2 — Module-level random bypasses the seeded RNG

model.py imported the stdlib random module and called random.sample() and random.uniform() during agent initialisation. This means the rng parameter passed to EmperorModel() — and exposed as a slider in the UI — had no effect on the initial believer placement or conviction values, making runs non-reproducible. Fixed by replacing both calls with self.random.sample() and self.random.uniform().

Test plan

• solara run app.py starts without ImportError
• Agents render on the grid with correct colors (comply/enforce states)
• Changing the Random Seed slider and clicking Reset produces a different initial configuration
• Line plot shows Compliance / Enforcement / False Enforcement over time

🤖 Generated with Claude Code

[sadrasabouri]

Hey @abhinavk0220
I verify that the issue you're trying to solve is still there (Python 3.12.13 and Mesa 4.0.0a0). I reviewed your PR, and here are some comments that can enhance it:

• Currently, in examples/emperor_dilemma/README.md we have solara run app.py as the run command. This implicitly assumes that the user is running the example when the working directory is examples/emperor_dilemma. Therefore, your fix will result in ModuleNotFoundError: No module named 'emperor_dilemma'. My suggestion is to move the agents.py and model.py into an insider directory emperor_dilemma (similar to bank_reserves).
• nit: make atomic-scoped PRs. The two bugs being solved by this PR do not live in the same scope and could be two separate valid PR.

The rest looks good to me.