Exercise 05: add symbolic Chow closure documentation, evaluation and visualization (public)

[paolobasso1211]

Changes

• Added the student-facing compute_lie_bracket and compute_chow_closure functions.
• Added public Task 7 problem instances in data.py, including regular and singular evaluation points for unicycle, bicycle, Martinet and non-controllable translation systems.
• Kept the evaluation point in problem_def.py outside the student API so that the closure must remain symbolic.
• Added the Task 7 evaluation flow:
  • the student implementation returns a symbolic closure matrix;
  • the evaluator substitutes the evaluation point;
  • the resulting matrix rank is compared with the expected ground truth.
• Added Task 7 scoring (5%) and reduced Task 4 (from 50% to 45%)
• Added symbolic visualization plots in problem_def.py and at the end of utils.py for:
  • the input vector fields;
  • the returned Chow closure matrix.
• Added utility functions compute_middle_curve and update_arc_length in utils.py and imported in algo.py to simplify Task 4 for students
• Updated the Exercise 05 documentation with the new task specification and API.
• Regenerated expected.pickle for the updated public evaluation cases.
• Updated the documentation 05-dubinspath.md to describe the symbolic-closure API and evaluation procedure.
• Cleaned ex05.py and added dashboard output management for matrix messages for Task 7

Validation

All public reference cases for Tasks 7 pass with accuracy 1.0.