Add server tool call support to OpenTelemetryChatClient per semantic conventions

[Copilot]

• Understand existing content types
• Create specialized server tool call and response classes per feedback
• Add ExtractCodeFromInputs helper for code_interpreter
• Update case handlers to use specialized classes
• Update JsonSerializable attributes for generic types
• Update tests to match new JSON output format
• Build and test the changes
• Use dc.HasTopLevelMediaType instead of StartsWith
• Remove redundant application/x-python-code condition
• Invert null check for single return null
• Remove unnecessary pragma warnings

Original prompt

Update the OpenTelemetryChatClient to accomodate the additions to the standard convention in open-telemetry/semantic-conventions@08de1bc. For example, CodeInterpreterToolCallContent and CodeInterpreterToolResultContent should be recognized and translated to server_tool_call and server_tool_call_response in the otel output, as should ImageGenerationToolCallContent, ImageGenerationResultCallContent, McpServerToolCallContent, McpServerToolResultContent, McpServerToolApprovalRequestContent, and McpServerToolApprovalResponseContent. Create the appropriate types for the otel ServerToolCallPart, ServerToolCallResponsePart, GenericServerToolCall, and GenericServerToolCallResponse to make it easier to accomodate additional server tools in the future.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

Microsoft Reviewers: Open in CodeFlow

[Copilot]

Pull request overview

This PR extends OpenTelemetryChatClient to support new OpenTelemetry semantic conventions for server tool calls, covering code interpreter, image generation, and MCP server tooling, and updates tests accordingly. It introduces specialized OTEL serialization types and a helper to extract code interpreter input, ensuring both streaming and non-streaming paths emit the expected OTEL JSON.

Changes:

• Map CodeInterpreterToolCallContent, CodeInterpreterToolResultContent, ImageGenerationToolCallContent, ImageGenerationToolResultContent, McpServerToolCallContent, McpServerToolResultContent, McpServerToolApprovalRequestContent, and McpServerToolApprovalResponseContent into OTEL server_tool_call / server_tool_call_response parts with dedicated internal DTOs.
• Add ExtractCodeFromInputs to derive the code field for code interpreter calls from AIContent inputs and register the new OTEL DTOs with JsonSerializable for source-generated serialization.
• Add tests to verify the new server tool call and MCP approval content types serialize to the expected OTEL JSON for both single-shot and streaming chat flows.

Reviewed changes

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

File	Description
src/Libraries/Microsoft.Extensions.AI/ChatCompletion/OpenTelemetryChatClient.cs	Adds server tool call / response part types, content-to-OTEL mapping for new server tool content types, the ExtractCodeFromInputs helper, and updates OTEL JSON source-generation registrations.
test/Libraries/Microsoft.Extensions.AI.Tests/ChatCompletion/OpenTelemetryChatClientTests.cs	Adds tests to assert correct OTEL JSON for server tool calls/results (code interpreter, image generation, MCP) and MCP approval request/response content, for both streaming and non-streaming scenarios.

Comments suppressed due to low confidence (1)

src/Libraries/Microsoft.Extensions.AI/ChatCompletion/OpenTelemetryChatClient.cs:493

• ExtractCodeFromInputs handles both DataContent (by decoding dc.Data when the media type is text/* or application/x-python-code) and TextContent, but the new tests exercise only the TextContent code path. To reduce the risk of regressions and to validate the primary DataContent-based scenario used by the OpenAI clients (which populate Inputs with DataContent using a "text/x-python" media type), consider adding a test that creates a CodeInterpreterToolCallContent with a DataContent input and verifies that the serialized server_tool_call.server_tool_call.code field contains the expected decoded code string.

    private static string? ExtractCodeFromInputs(IList<AIContent>? inputs)
    {
        if (inputs is not null)
        {
            foreach (var input in inputs)
            {
                // Check for DataContent with text MIME types
                if (input is DataContent dc && dc.HasTopLevelMediaType("text"))
                {
                    // Return the data as a string (decode bytes as UTF8)
                    return Encoding.UTF8.GetString(dc.Data.ToArray());
                }

                // Check for TextContent
                if (input is TextContent tc && !string.IsNullOrEmpty(tc.Text))
                {
                    return tc.Text;
                }
            }
        }

        return null;
    }

    /// <summary>Creates an activity for a chat request, or returns <see langword="null"/> if not enabled.</summary>
    private Activity? CreateAndConfigureActivity(ChatOptions? options)
    {
        Activity? activity = null;