Add support for minimal reasoning effort

[SeeknnDestroy]

Fixes #1407

Summary

• depend on OpenAI Python 1.99.3 to expose new minimal reasoning effort
• adjust converters and LiteLLM integration for updated OpenAI tool-call types
• cover reasoning effort in model settings and add regression test

Testing

• make format
• make lint
• make mypy
• make tests

[serialx]

I believe 0.2.6 already implements GPT-5 related features.

[seratch]

Yes, you're right. You can pass reasoning effort this way:

from agents import Agent, Runner, ModelSettings
from openai.types.shared import Reasoning

agent = Agent(
    name="Assistant",
    instructions="You are a helpful assistant",
    model="gpt-5",
    model_settings=ModelSettings(
        reasoning=Reasoning(effort="minimal"),
    ),
)

result = Runner.run_sync(agent, "Write a haiku about recursion in programming.")
print(result.final_output)

Thanks again for taking the time for this PR, and let us close this now.