fix(agents): accept generation kwargs on LlmAgent - #6837

Summary

• LlmAgent(temperature=0.1) (and other GenerateContentConfig knobs such as top_p and max_output_tokens) now folds into generate_content_config instead of failing with extra_forbidden.
• Reserved knobs that already have an agent field (system_instruction, response_schema) raise a message that points at instruction= / output_schema=; setting the same field both as a kwarg and inside generate_content_config is rejected.
• The LlmAgent class and generate_content_config field docstrings now document the mapping.

Closes: #6836

Test plan

• I have added or updated unit tests for my change.
• All unit tests pass locally.

Unit Tests:

uv run python -m pytest tests/unittests/agents/test_llm_agent_fields.py tests/unittests/agents/test_llm_agent_error_messages.py -q
107 passed

Covers the original crash (temperature= with output_schema), merging with an existing generate_content_config, camelCase aliases, thinking_config kwargs, reserved-field redirects, conflicts, and unknown extras remaining forbidden.

Manual End-to-End (E2E) Tests:

from google.adk.agents import LlmAgent
from pydantic import BaseModel

class MySchema(BaseModel):
    score: int

agent = LlmAgent(
    name="grader",
    model="gemini-3.5-flash-lite",
    instruction="Grade the exam.",
    output_schema=MySchema,
    temperature=0.1,
)
assert agent.generate_content_config.temperature == 0.1

Previously this raised ValidationError: temperature Extra inputs are not permitted.

Checklist

• I have read the CONTRIBUTING.md document.
• I have performed a self-review of my own code.
• I have added tests that prove my fix is effective or that my feature works.
• New and existing unit tests pass locally with my changes.