fix: preserve Gemini thinking blocks

[he-yufeng]

Summary

• recognize Gemini-shaped thinking_blocks even when blocks do not carry Anthropic-style per-block signatures
• merge LiteLLM's parallel provider_specific_fields.thought_signatures into those blocks before converting them to ADK thought parts
• add regression coverage for Gemini blocks with and without signatures, plus the generated response conversion path

Fixes #5712.

To verify

• python -m pytest tests/unittests/models/test_litellm.py -q -k "thinking_blocks or thought_signature" --basetemp .tmp/pytest
• python -m pytest tests/unittests/models/test_litellm.py -q --basetemp .tmp/pytest
• python -m py_compile src/google/adk/models/lite_llm.py tests/unittests/models/test_litellm.py
• python -m pyink --check src/google/adk/models/lite_llm.py tests/unittests/models/test_litellm.py

[he-yufeng]

Rebased this onto current upstream/main and force-pushed the branch. The PR diff is still limited to the LiteLLM Gemini thinking-block handling and its unit coverage.

Local validation after the rebase:

• .venv\Scripts\python.exe -m pytest tests/unittests/models/test_litellm.py -q -k "thinking_blocks or thought_signature" (18 passed)
• .venv\Scripts\python.exe -m py_compile src/google/adk/models/lite_llm.py tests/unittests/models/test_litellm.py
• .venv\Scripts\pyink.exe --check src/google/adk/models/lite_llm.py tests/unittests/models/test_litellm.py
• .venv\Scripts\isort.exe --check-only src/google/adk/models/lite_llm.py tests/unittests/models/test_litellm.py
• git diff --check upstream/main..HEAD

The remaining visible failure is still the repository triage bot, not a code test failure.

[rohityan]

Hi @he-yufeng , Thank you for your contribution! We appreciate you taking the time to submit this pull request. Your PR has been received by the team and is currently under review. We will provide feedback as soon as we have an update to share.

[rohityan]

Hi @sasha-gitg , can you please review this.