fix(openai): normalize Pydantic objects before JSON serialization in truncate_messages_by_size

[mecampbellsoup]

Summary

The OpenAI SDK v1+ returns Pydantic model instances (e.g., ResponseFunctionToolCall) which are not directly JSON serializable. When these objects are passed to truncate_messages_by_size(), the json.dumps() call fails with a TypeError:

TypeError: Object of type ResponseFunctionToolCall is not JSON serializable

The _normalize_data() helper already exists in this file and properly handles Pydantic objects by calling .model_dump(), but it wasn't being used in truncate_messages_by_size().

The Fix

This PR adds normalization at the start of truncate_messages_by_size() to ensure all Pydantic objects are converted to JSON-compatible dicts before serialization:

# Normalize messages to ensure JSON serialization works
# (handles Pydantic objects from OpenAI SDK v1+)
messages = _normalize_data(messages, unpack=False)

Reproduction

When using the OpenAI SDK's Responses API with tool calls and Sentry's OpenAI integration enabled:

from openai import OpenAI
from openai.types.responses import ResponseFunctionToolCall

client = OpenAI()
response = client.responses.create(
    model="gpt-4o",
    input=[
        {"role": "user", "content": "What's the weather?"},
        {
            "role": "assistant", 
            "content": [ResponseFunctionToolCall(
                type="function_call",
                id="call_123",
                call_id="call_123",
                name="get_weather",
                arguments='{"location": "NYC"}'
            )]
        }
    ]
)

This crashes with the TypeError because Sentry's integration tries to serialize the messages for tracing.

Fixes #5350

[github-actions]

Semver Impact of This PR

🟢 Patch (bug fixes)

📋 Changelog Preview

This is how your changes will appear in the changelog.
Entries from this PR are highlighted with a left border (blockquote style).

---

Bug Fixes 🐛

• fix(openai): normalize Pydantic objects before JSON serialization in truncate_messages_by_size by mecampbellsoup in #5351

Internal Changes 🔧

• ci: Fix path in AI integration tests by alexander-alderman-webb in #5347

---

_{🤖 This preview updates automatically when you update the PR.}

[alexander-alderman-webb]

Hi @mecampbellsoup,

Thanks for raising the bug, we're adjusting the trimming logic for the next release, so there will not be any serialization in the integrations any longer. See #5335 for details.