Refactor Noma guardrail to use shared Responses transformation and include system instructions

[idola9]

Title

Refactor Noma guardrail to use shared Responses transformation and include system instructions

Relevant issues

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/litellm/ directory, Adding at least 1 test is a hard requirement - see details
• I have added a screenshot of my new test passing locally
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem

Type

🆕 New Feature
🧹 Refactoring
✅ Test

Changes

• Aggregate system prompts in the Responses transformer itself: convert_chat_completion_messages_to_responses_api no longer overwrites instructions on each string system message; it now concatenates multiple system prompts into a single string (instructions = f"{instructions} {content}" when instructions already exists), which is then surfaced through transform_request.
• Use shared Responses transformer in Noma guardrail: NomaGuardrail now uses LiteLLMResponsesTransformationHandler.convert_chat_completion_messages_to_responses_api instead of custom message-extraction helpers, and builds the Noma payload from input_items.
• Forward system prompts to Noma as a single instructions message: multiple string system messages are concatenated into instructions, then turned into a single system message prepended to input before calling the Noma API; tests for single and multiple system prompts were updated accordingly.
• Reuse transformer for image and mixed content tests: the test_extract_user_message_* and test_image_with_base64_data tests now call the transformer directly, asserting the same shapes as before (text + input_image blocks, including base64 URLs) using AllMessageValues and proper typing.

[vercel]

@idola9 is attempting to deploy a commit to the CLERKIEAI Team on Vercel.

A member of the Team first needs to authorize it.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}