Add a runnable example for a taxonomy task that lacks one

[raphaelrrcoelho]

The Tour promises "a runnable example for every row," and examples/ already covers MABUC, CGFA, sensitivity bounds, offline→online, and where-to-intervene. Add one more in the same style as examples/sensitivity_bounds.py (module docstring, main(), __main__ guard) for a task with none yet — e.g. reward shaping (causal_potential, q_learning) or causal curricula (causal_curriculum). CPU-only, < ~40 lines.

Files: examples/<task>_example.py.