A Comprehensive Mechanistic Interpretability Explainer & Glossary — LessWrong