(ai-gateway): Create how-to for guarding LLM responses using the OpenAI Moderation API

The AI Custom Guardrail plugin currently has one how-to guide covering the Mistral AI Moderation API. The plugin supports any HTTP-based guardrail service, single-direction guarding, and metrics logging, none of which are covered by a dedicated guide. This issue tracks three guides to fill that gap.

What it covers:

• Configure the AI Custom Guardrail plugin with guarding_mode: OUTPUT to inspect only LLM responses - this might need some testing or changing, as guarding for responses proved to be not fully reliable for other guardrail plkugins
• Use the OpenAI Moderation API (/v1/moderations) as the guardrail service
• Write a check_response function that evaluates results[0].flagged and uses results[0].category_scores to identify which categories triggered
• Validate that a clean request passes and a response containing harmful content is blocked
• Add logging plugin to the how-to to show how blocked responses are logged

Size: S

Start by reading the existing Mistral AI Moderation API how-to and the AI Custom Guardrail plugin documentation. Test OUTPUT guarding with a clean request and harmful response, then document the OpenAI /v1/moderations call, check_response behavior, category logging, and blocked-response logging. Done means the three guides cover configuration, validation, and logging.