A Three-Branch Checks-and-Balances Frameworkfor Context-Aware Ethical Alignment of Large Language Models