feat(gui): Implement per-response token metrics and AI discussion compression

- Display token metrics (input/output/cache) per response in Discussion Hub.
- Add total Discussion Token usage in the panel header.
- Implement 'Compress' feature to intelligently summarize and replace exhausted discussion histories using an AI subagent.

src/app_controller.py

@@ -2323,6 +2323,8 @@ class AppController:
       "collapsed": True,
       "ts": entry.get("ts", project_manager.now_ts())
      }
+     if "usage" in payload:
+      entry_obj["usage"] = payload["usage"]
      if segments:
       entry_obj["thinking_segments"] = [{"content": s.content, "marker": s.marker} for s in segments]
 
@@ -3259,6 +3261,26 @@ class AppController:
     self.ai_status = f"error: {e}"
   threading.Thread(target=worker, daemon=True).start()
 
+ def _handle_compress_discussion(self) -> None:
+  def worker():
+   try:
+    self.ai_status = "compressing discussion..."
+    disc_text = project_manager.format_discussion(self.disc_entries)
+    if not disc_text.strip():
+     self.ai_status = "discussion is empty"
+     return
+    
+    response_text = ai_client.run_discussion_compression(disc_text)
+    
+    if response_text and not response_text.startswith("ERROR:"):
+     self.disc_entries = [{"role": "System", "content": f"[COMPRESSED HISTORY]\n{response_text}", "collapsed": False, "ts": project_manager.now_ts()}]
+     self.ai_status = "compression complete"
+    else:
+     self.ai_status = f"compression failed: {response_text}"
+   except Exception as e:
+    self.ai_status = f"compression error: {e}"
+  threading.Thread(target=worker, daemon=True).start()
+
  def _handle_generate_send(self) -> None:
   """