Beyond Compromise: Pareto-Lenient Consensus for Efficient Multi-Preference LLM Alignment | ScienceToStartup