Implement Anonymizers and Guardrails Pipelines in `endpoint_i

[radlab-dev]

Description

Enhance the endpoint_i endpoint to incorporate dedicated processing pipelines for data anonymization and guardrails enforcement. The guardrails logic must be split into two separate pipelines that run before and after the main request handling.

Tasks

• Add Anonymizers Pipeline

  • Integrate a pipeline that applies anonymization steps to incoming data prior to core business logic.
  • Make the pipeline configurable (enable/disable specific anonymizers, ordering, etc.).

• Add Guardrails Pipelines

  • Before‑request pipeline
    • Implement a guardrails pipeline that validates the request (e.g., policy checks, PII detection) before the main processing.
  • After‑request pipeline
    • Implement a guardrails pipeline that validates or sanitizes the response after the main processing.
  • Allow independent configuration for each guardrails pipeline (enable/disable, custom rules, thresholds).

• Wire Pipelines into endpoint_i

  • Ensure the execution order is:
    1. before_request_pipeline (guardrails)
    2. Anonymizers pipeline
    3. Core request handling
    4. after_request_pipeline (guardrails)

• Documentation

  • Update the README and API documentation with usage examples, configuration options, and instructions for enabling/disabling each pipeline.