[opt_transport] shuffle example output no longer matches description

In opt_transport.md, the constraint-shuffling example currently gives only one distinct transportation plan, but the text says it “uncovers two optimal transportation plans.”

The LP does have multiple optimal solutions. However, with the current SciPy/HiGHS solver, shuffling the order of the equality constraints does not appear to return a different optimal plan.

I think we should replace or remove the shuffle example.

One possible replacement is to solve the LP, record the optimal value, add the equality constraint C_vec @ x == optimal_value, and then solve secondary LPs over the optimal face to exhibit other optimal plans with the same objective value.

I think it would be very useful for the lecture to show a deterministic way to find multiple optimal plans, rather than relying on random shuffling or solver tie-breaking!

CC @mmcky and @Chihiro2000GitHub and referencing #892.

Start in opt_transport.md by reading and running the constraint-shuffling example with the current SciPy/HiGHS solver. Replace or remove the shuffle-based demonstration, using the proposed optimal-face approach if appropriate, and ensure the lecture deterministically shows multiple optimal transportation plans with the same objective value.