AI: add enable_thinking reasoning toggle plumbed to llama.cpp

New optional SamplingOverride forwarded to llama-server as
chat_template_kwargs.enable_thinking (gates Qwen3-style reasoning
blocks). None leaves the template default; other backends ignore it.
Wired through the agentic-insight and chat-turn request bodies/handlers.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

src/ai/handlers.rs

@@ -40,6 +40,12 @@ pub struct GeneratePhotoInsightRequest {
     pub top_k: Option<i32>,
     #[serde(default)]
     pub min_p: Option<f32>,
+    /// Reasoning toggle for thinking-capable models. Forwarded to the
+    /// llama.cpp backend as `chat_template_kwargs.enable_thinking`; ignored
+    /// by other backends and the non-agentic (Ollama) path. Only the agentic
+    /// endpoint routes through llama.cpp. None defers to the template default.
+    #[serde(default)]
+    pub enable_thinking: Option<bool>,
     /// `"local"` (default, Ollama with images) | `"hybrid"` (local vision +
     /// OpenRouter chat). Only respected by the agentic endpoint.
     #[serde(default)]
@@ -868,6 +874,7 @@ pub async fn generate_agentic_insight_handler(
                     request.top_p,
                     request.top_k,
                     request.min_p,
+                    request.enable_thinking,
                     max_iterations,
                     request.backend.clone(),
                     fewshot_examples,
@@ -1169,6 +1176,11 @@ pub struct ChatTurnHttpRequest {
     pub top_k: Option<i32>,
     #[serde(default)]
     pub min_p: Option<f32>,
+    /// Reasoning toggle for thinking-capable models. Forwarded to the
+    /// llama.cpp backend as `chat_template_kwargs.enable_thinking`; ignored
+    /// by other backends. None defers to the model/template default.
+    #[serde(default)]
+    pub enable_thinking: Option<bool>,
     #[serde(default)]
     pub max_iterations: Option<usize>,
     /// Per-turn system-prompt override. Ephemeral in append mode,
@@ -1247,6 +1259,7 @@ pub async fn chat_turn_handler(
         top_p: request.top_p,
         top_k: request.top_k,
         min_p: request.min_p,
+        enable_thinking: request.enable_thinking,
         max_iterations: request.max_iterations,
         system_prompt: request.system_prompt.clone(),
         persona_id: request.persona_id.clone(),
@@ -1473,6 +1486,7 @@ pub async fn chat_stream_handler(
         top_p: request.top_p,
         top_k: request.top_k,
         min_p: request.min_p,
+        enable_thinking: request.enable_thinking,
         max_iterations: request.max_iterations,
         system_prompt: request.system_prompt.clone(),
         persona_id: request.persona_id.clone(),
@@ -1618,6 +1632,7 @@ pub async fn turn_async_handler(
         top_p: request.top_p,
         top_k: request.top_k,
         min_p: request.min_p,
+        enable_thinking: request.enable_thinking,
         max_iterations: request.max_iterations,
         system_prompt: request.system_prompt.clone(),
         persona_id: request.persona_id.clone(),