Python: Fix Gemini harness tool declarations

[scarab-systems]

Motivation & Context

Fixes #6954.

Gemini Developer API calls can combine native Gemini tools, such as Google Search grounding, with Agent Framework function tools. That mixed-tool path needs Gemini's server-side tool invocation reporting enabled so native tool calls are returned alongside function calls.

Function tool schemas also need to be forwarded through the Gemini SDK's JSON Schema field so richer Agent Framework tool schemas are preserved instead of being coerced into the SDK Schema subset.

Disclosure: This PR was prepared with AI assistance under human direction and review.

Description & Review Guide

• What are the major changes?
  • Pass FunctionTool schemas through parameters_json_schema when building Gemini function declarations.
  • Enable include_server_side_tool_invocations for Gemini Developer API requests that mix native Gemini tools with function declarations.
  • Preserve existing tool_choice / function-calling configuration when adding that Developer API flag.
  • Add Gemini client tests for JSON Schema pass-through, mixed native/function tool config, preserved tool_choice, and Vertex AI behavior.
• What is the impact of these changes?
  • The failing Gemini harness mixed-tool scenario gets the native tool-call reporting it needs while keeping Vertex AI behavior unchanged.
• What do you want reviewers to focus on?
  • Whether the Developer API-only condition is the right scope for enabling server-side tool invocation reporting on mixed native/function tool requests.

Related Issue

Fixes #6954

Contribution Checklist

• The code builds clean without any errors or warnings
• All unit tests pass, and I have added new tests where possible
• The PR follows the Contribution Guidelines
• This PR is linked to an issue and there is no other open PR for this issue (see Related Issue above).
• This is not a breaking change. If it is a breaking change, add the breaking change label (or add "[BREAKING]" to the title prefix, before or after any language prefix) -- a workflow keeps the label and title prefix in sync automatically.

[Copilot]

Pull request overview

This PR fixes Gemini (Developer API) tool-calling interoperability in the Python Gemini client so harness scenarios can combine native Gemini tools (e.g., Google Search grounding) with Agent Framework FunctionTools without request-prep validation failures or Gemini API rejections.

Changes:

• Forward full FunctionTool JSON Schema via types.FunctionDeclaration.parameters_json_schema (instead of coercing into the SDK’s parameters Schema subset).
• When using the Gemini Developer API and the request mixes native Gemini tools with function declarations, set tool_config.include_server_side_tool_invocations = True while preserving any existing function-calling configuration (tool_choice).
• Add/extend Gemini client tests covering JSON Schema pass-through, mixed native/function tools behavior, tool_choice preservation, and Vertex AI non-impact.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
python/packages/gemini/agent_framework_gemini/_chat_client.py	Updates tool declaration building to use JSON Schema passthrough and conditionally enables Developer API server-side tool invocation reporting for mixed native/function tool requests.
python/packages/gemini/tests/test_gemini_client.py	Adds tests verifying JSON Schema forwarding, server-side tool invocation flag behavior for Developer API mixed-tool requests, tool_choice preservation, and unchanged Vertex AI behavior.