fix(bleed): fix token budget layout — own collapsing header in AI Settings

Phase 3 changes:
- _render_provider_panel: removed 4 redundant lines (2x 'Token Budget' labels,
  separator, embedded _render_token_budget_panel call)
- _gui_func AI Settings: added collapsing_header('Token Budget') section after
  'System Prompts', calling _render_token_budget_panel cleanly
AI Settings now has three independent collapsing sections.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

gui_2.py

@@ -1695,6 +1695,8 @@ class App:
       self._render_provider_panel()
      if imgui.collapsing_header("System Prompts"):
       self._render_system_prompts_panel()
+     if imgui.collapsing_header("Token Budget"):
+      self._render_token_budget_panel()
     imgui.end()
    if self.show_windows.get("MMA Dashboard", False):
     exp, self.show_windows["MMA Dashboard"] = imgui.begin("MMA Dashboard", self.show_windows["MMA Dashboard"])
@@ -2713,10 +2715,6 @@ class App:
    imgui.text_colored(C_LBL, f"  Last Latency: {usage['last_latency']:.2f}s")
   if usage["cache_read_input_tokens"]:
    imgui.text_colored(C_LBL, f"  Cache Read: {usage['cache_read_input_tokens']:,}  Creation: {usage['cache_creation_input_tokens']:,}")
-  imgui.text("Token Budget:")
-  imgui.separator()
-  imgui.text("Token Budget")
-  self._render_token_budget_panel()
   if self._gemini_cache_text:
    imgui.text_colored(C_SUB, self._gemini_cache_text)