Important: finalize_hackathon insertion sort is O(n²)

[0xdevcollins]

Severity: Important (performance)

Current sort:

// contract.rs:474
let mut inserted = false;
for j in 0..leads.len() {
    if avg > scores.get(j).unwrap() {
        leads.insert(j, lead.clone());   // O(n) per insert
        scores.insert(j, avg);
        inserted = true;
        break;
    }
}

For 100 submissions: ~10,000 Vec operations. For 1,000: 1M ops. Soroban resource fees scale with this — finalize for a popular hackathon could become prohibitively expensive or fail resource limits.

Fix

Collect into Vec, then use sort_by once with explicit tiebreaker (see related issue on stable sort).

Tests

• Benchmark 100 submissions before/after
• Verify ordering matches previous behavior on small inputs