Check text for profanity, toxicity, sentiment, and language. Returns per-category flags and an is_safe boolean for gating content.
