Refactor CLI and LLM runtime handling for improved user experience

- Removed deprecated console usage in favor of structured logging functions for error handling and user prompts.
- Enhanced CLI prompts for LLM runtime and preset selection, utilizing menu helpers for better user interaction.
- Updated GPU pool scanning output with improved formatting and error indication for clarity.
- Refactored setup wizard to streamline LLM runtime and preset configuration, ensuring a more intuitive setup process.
- Improved documentation and user feedback in CLI outputs to enhance overall usability.

src/gpu_rent/llm_runtime.py

@@ -26,12 +26,18 @@ OLLAMA_PRESETS: dict[str, list[str]] = {
     "empty": [],
 }
 
-PRESET_HELP = (
-    "recommended — Qwen2.5 7B abliterate (RU/EN, мало отказов, ~5GB)\n"
-    "light — qwen2.5:3b (быстрее, слабее)\n"
-    "stock — официальный qwen2.5:7b (больше цензуры)\n"
-    "alt — другой abliterate-пак 7B\n"
-    "empty — только runtime, без pull"
+OLLAMA_PRESET_LABELS: dict[str, str] = {
+    "recommended": "Qwen2.5 7B abliterate (RU/EN, мало отказов, ~5GB)",
+    "light": "qwen2.5:3b (быстрее, слабее)",
+    "stock": "официальный qwen2.5:7b (больше цензуры)",
+    "alt": "другой abliterate-пак 7B",
+    "empty": "только runtime, без pull",
+    "keep": "не менять ollama-models.yaml",
+}
+
+# Deprecated text blob — prefer menu helpers below.
+PRESET_HELP = "\n".join(
+    f"{k} — {v}" for k, v in OLLAMA_PRESET_LABELS.items() if k != "keep"
 )
 
 LLAMACPP_PRESETS: dict[str, list[str]] = {
@@ -47,13 +53,48 @@ LLAMACPP_PRESETS: dict[str, list[str]] = {
     "empty": [],
 }
 
-LLAMACPP_PRESET_HELP = (
-    "recommended — Qwen2.5 7B abliterate GGUF Q4_K_M (~4.7GB, мало отказов)\n"
-    "light — Qwen2.5 3B Instruct Q4_K_M (~2GB)\n"
-    "stock — официальный Qwen2.5 7B Instruct Q4_K_M\n"
-    "empty — только llama-server, GGUF положи вручную / правь llamacpp-models.yaml"
+LLAMACPP_PRESET_LABELS: dict[str, str] = {
+    "recommended": "Qwen2.5 7B abliterate GGUF Q4_K_M (~4.7GB, мало отказов)",
+    "light": "Qwen2.5 3B Instruct Q4_K_M (~2GB)",
+    "stock": "официальный Qwen2.5 7B Instruct Q4_K_M",
+    "empty": "только llama-server, GGUF вручную",
+    "keep": "не менять llamacpp-models.yaml",
+}
+
+LLAMACPP_PRESET_HELP = "\n".join(
+    f"{k} — {v}" for k, v in LLAMACPP_PRESET_LABELS.items() if k != "keep"
 )
 
+LLM_RUNTIME_LABELS: dict[str, str] = {
+    "none": "только SwarmUI",
+    "ollama": "Ollama (+ pull моделей)",
+    "llamacpp": "llama.cpp server (+ GGUF)",
+}
+
+
+def llm_runtime_menu() -> list:
+    from gpu_rent.prompts import MenuItem
+
+    return [MenuItem(k, f"{k} — {LLM_RUNTIME_LABELS[k]}") for k in ("none", "ollama", "llamacpp")]
+
+
+def ollama_preset_menu(*, include_keep: bool = False) -> list:
+    from gpu_rent.prompts import MenuItem
+
+    keys = list(OLLAMA_PRESETS.keys())
+    if include_keep:
+        keys.append("keep")
+    return [MenuItem(k, OLLAMA_PRESET_LABELS.get(k, k)) for k in keys]
+
+
+def llamacpp_preset_menu(*, include_keep: bool = False) -> list:
+    from gpu_rent.prompts import MenuItem
+
+    keys = list(LLAMACPP_PRESETS.keys())
+    if include_keep:
+        keys.append("keep")
+    return [MenuItem(k, LLAMACPP_PRESET_LABELS.get(k, k)) for k in keys]
+
 
 @dataclass(frozen=True)
 class OllamaModelEntry: