fix: tolerate partial LiteLLM citations

[devin-lai]

Summary

Fix LiteLLM citation annotation conversion so provider-shaped partial URL citations do not raise KeyError or produce unusable empty citations. The converter now preserves valid citation data, defaults missing title and index fields safely, skips entries without citation URLs, and returns None when no usable citations remain.

Test plan

• uv run pytest tests/models/test_litellm_annotations.py
• PATH="/usr/bin:/bin:/usr/sbin:/sbin:/Users/ldy/.local/bin:/Users/ldy/miniforge3/bin:/opt/homebrew/bin" bash .agents/skills/code-change-verification/scripts/run.sh

Issue number

N/A

Checks

• I've added new tests (if relevant)
• I've added/updated the relevant documentation
• I've run make lint and make format
• I've made sure tests pass

[github-actions]

This PR is stale because it has been open for 10 days with no activity.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 0a3942c739

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Propagate converted annotations to run output

When LiteLLM returns a valid message.annotations list, this helper now builds Annotation objects, but the non-streaming LitellmModel.get_response path immediately passes the message into Converter.message_to_output_items, which still creates ResponseOutputText(..., annotations=[]) in src/agents/models/chatcmpl_converter.py. As a result, users of Runner.run still receive empty annotations even though this converter accepted the citation, and the new tests miss that end-to-end path.

Useful? React with 👍 / 👎.