Add Gini coefficient, Kurtosis, Percentile rank, and Atkinson index to unsorted module

[Copilot]

Adds four inequality and distribution shape statistics to the unsorted module, completing the statistical analysis toolkit for streaming data.

New Functions

• gini() - Gini coefficient for measuring inequality (0 = perfect equality, 1 = perfect inequality)
• kurtosis() - Excess kurtosis for measuring distribution tailedness (0 = normal, positive = heavy tails, negative = light tails)
• percentile_rank() - Percentile rank of a value within a distribution (0-100 scale)
• atkinson() - Atkinson index with configurable inequality aversion parameter ε

Implementation Details

• All functions use O(n log n) sorting with O(n) space complexity
• Adaptive parallel processing: sequential for <10k elements, parallel for ≥10k
• Proper edge case handling: empty data, single elements, zero variance, zero sums
• Binary search optimization in percentile_rank() for efficient lookups
• Special-case geometric mean computation in atkinson() when ε=1

Usage

use stats::unsorted::{gini, kurtosis, percentile_rank, atkinson};

// Measure income inequality
let incomes = vec![20_000, 30_000, 45_000, 80_000, 200_000];
let inequality = gini(incomes.iter().copied()).unwrap();  // ~0.34

// Check distribution shape
let returns = vec![-2.1, -0.5, 0.3, 0.8, 1.2, 5.7];
let tail_heaviness = kurtosis(returns.iter().copied()).unwrap();

// Find relative position
let score_rank = percentile_rank(scores.iter().copied(), 85).unwrap();

// Inequality with aversion to low-end gaps (ε=1.5)
let adjusted_inequality = atkinson(incomes.iter().copied(), 1.5).unwrap();

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[jqnatividad]

Thanks for your review @copilot