agent: add sampling parameters (temperature, top_p, top_k)

Move temperature from a per-call parameter to an Agent field,
add top_p and top_k. All three are sent to the API via a new
SamplingParams struct, displayed on the F5 thalamus screen.

Defaults: temperature=0.6, top_p=0.95, top_k=20 (Qwen3.5 defaults).

Also adds top_p and top_k to ChatRequest so they're sent in the
API payload. Previously only temperature was sent.

UI controls for adjusting these at runtime are not yet implemented.

Co-Authored-By: Proof of Concept <poc@bcachefs.org>

src/agent/api/openai.rs

@@ -26,7 +26,7 @@ pub(super) async fn stream_events(
     tx: &mpsc::UnboundedSender<StreamEvent>,
     ui_tx: &UiSender,
     reasoning_effort: &str,
-    temperature: Option<f32>,
+    sampling: super::SamplingParams,
     priority: Option<i32>,
 ) -> Result<()> {
     let request = ChatRequest {
@@ -35,7 +35,9 @@ pub(super) async fn stream_events(
         tool_choice: tools.map(|_| "auto".to_string()),
         tools: tools.map(|t| t.to_vec()),
         max_tokens: Some(16384),
-        temperature: Some(temperature.unwrap_or(0.6)),
+        temperature: Some(sampling.temperature),
+        top_p: Some(sampling.top_p),
+        top_k: Some(sampling.top_k),
         stream: Some(true),
         reasoning: if reasoning_effort != "none" && reasoning_effort != "default" {
             Some(ReasoningConfig {