123. Cohen's Kappa

Implement Cohen's kappa, an inter-rater agreement metric that corrects for chance agreement.

Signature: def cohen_kappa(rater1: list, rater2: list) -> float

Where:

• p_o = observed agreement = fraction of items both raters agree on
• p_e = expected agreement by chance, computed from the marginals (sum_k p1_k * p2_k)
• kappa = (p_o - p_e) / (1 - p_e)

Returns 1.0 for perfect agreement, 0.0 for chance-level, negative for worse-than-chance.