feat(ai-core): wire function calling into the LLM pipeline

- llm: ChatStreamWithTools sends OpenAI-format tools; streamed tool_call
  deltas are reassembled by index into complete ToolCalls on the Done token
- model: LLMMessage gains ToolCallID/ToolCalls; StreamToken gains ToolCalls
- orchestrator: new ToolExecutor interface and streamWithTools loop —
  execute requested tools, feed results back, up to 3 rounds, final round
  forces a plain text answer; normal chats still stream token by token.
  Also adds Pipeline.FinishRound (assistant reply recording + post-round
  hook), wired to transports in the next commit.
- main: inject the plugin registry via toolExecAdapter

Includes an httptest-based test covering delta reassembly, tool
execution, and result feedback.

backend/services/ai-core/model/model.go

@@ -42,6 +42,11 @@ type LLMMessage struct {
 	Role    string `json:"role"`    // "system" | "user" | "assistant" | "tool"
 	Content string `json:"content"`
 	Name    string `json:"name,omitempty"`
+	// ToolCallID links a tool result message to the call that produced it (role="tool").
+	ToolCallID string `json:"tool_call_id,omitempty"`
+	// ToolCalls carries the calls requested by the assistant (role="assistant").
+	// Marshaled to the OpenAI wire format by the llm package, hence "-".
+	ToolCalls []ToolCall `json:"-"`
 }
 
 // LLMResponse represents a non-streaming LLM response.
@@ -54,9 +59,11 @@ type LLMResponse struct {
 
 // StreamToken is a single token from a streaming LLM response.
 type StreamToken struct {
-	Content      string `json:"content"`
-	FinishReason string `json:"finish_reason,omitempty"`
-	Done         bool   `json:"done"`
+	Content      string     `json:"content"`
+	FinishReason string     `json:"finish_reason,omitempty"`
+	Done         bool       `json:"done"`
+	// ToolCalls is set on the final Done token when the model requested tool calls.
+	ToolCalls    []ToolCall `json:"tool_calls,omitempty"`
 }
 
 // ReviewSegment represents a parsed action/chat segment from LLM output.