Litellm noma guardrail support images

[dean-zavad]

Guardrail Image Support

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
🐛 Bug Fix

Changes

• Migrating to noma v2 API for additional support in guardrailing images.
• Fixing partial output schema when integration with external logger

Notes

We removed apply_guardrails hook from noma code from several reasons:

• It broke datadog integration - this seems to be a bug beyond the scope of the noma guardrail.
• It broke guardrail features - guardrail "anonymize_input" and "monitor_mode" have stopped working.
• Streaming did not worked properly.

Once the above issues will be resolved, we'll be happy to opt in

Screenshot Of Passing Tests

[vercel]

@dean-zavad is attempting to deploy a commit to the CLERKIEAI Team on Vercel.

A member of the Team first needs to authorize it.

[CLAassistant]

All committers have signed the CLA.

[krrishdholakia]

@dean-zavad can you add apply_guardrail support in this PR - it would allow noma to work across all LLM endpoints (/v1/responses, /v1/messages, /audio/transcription, etc.)

[itaimodi]

we will implement that in another PR. apply_guardrail broke the current implementation. @ishaan-jaff