fix: preserve reasoning_content for thinking models in multi-turn conversations

GLM-Z1 and other thinking models require reasoning_content to be preserved
across tool call iterations. Without this, the API returns:
  'thinking is enabled but reasoning_content is missing in assistant tool call message'

Changes:
- Add ReasoningContent field to LLMResponse and Message types
- Parse reasoning_content from API response
- Preserve ReasoningContent when building assistant messages with tool calls

pkg/agent/loop.go

@@ -482,6 +482,7 @@ func (al *AgentLoop) runLLMIteration(ctx context.Context, messages []providers.M
 		assistantMsg := providers.Message{
 			Role:             "assistant",
 			Content:          response.Content,
+			ReasoningContent: response.ReasoningContent, // Preserve for thinking models (e.g., GLM-Z1)
 		}
 		for _, tc := range response.ToolCalls {
 			argumentsJSON, _ := json.Marshal(tc.Arguments)

pkg/providers/http_provider.go

@@ -127,6 +127,7 @@ func (p *HTTPProvider) parseResponse(body []byte) (*LLMResponse, error) {
 		Choices []struct {
 			Message struct {
 				Content          string `json:"content"`
+				ReasoningContent string `json:"reasoning_content"` // For thinking models (e.g., GLM-Z1)
 				ToolCalls []struct {
 					ID       string `json:"id"`
 					Type     string `json:"type"`
@@ -186,6 +187,7 @@ func (p *HTTPProvider) parseResponse(body []byte) (*LLMResponse, error) {
 
 	return &LLMResponse{
 		Content:          choice.Message.Content,
+		ReasoningContent: choice.Message.ReasoningContent,
 		ToolCalls:        toolCalls,
 		FinishReason:     choice.FinishReason,
 		Usage:            apiResponse.Usage,

pkg/providers/types.go

@@ -17,6 +17,7 @@ type FunctionCall struct {
 
 type LLMResponse struct {
 	Content          string     `json:"content"`
+	ReasoningContent string     `json:"reasoning_content,omitempty"` // For thinking models (e.g., GLM-Z1)
 	ToolCalls        []ToolCall `json:"tool_calls,omitempty"`
 	FinishReason     string     `json:"finish_reason"`
 	Usage            *UsageInfo `json:"usage,omitempty"`
@@ -31,6 +32,7 @@ type UsageInfo struct {
 type Message struct {
 	Role             string     `json:"role"`
 	Content          string     `json:"content"`
+	ReasoningContent string     `json:"reasoning_content,omitempty"` // For thinking models (e.g., GLM-Z1)
 	ToolCalls        []ToolCall `json:"tool_calls,omitempty"`
 	ToolCallID       string     `json:"tool_call_id,omitempty"`
 }