Refactor backend status handling and improve idle management

- Updated the idle-killer logic to treat SwarmUI `empty` and `disabled` states as busy, preventing unnecessary idle time during provisioning.
- Enhanced the `wait_backend_idle` function to recognize suspended backends as ready, improving resource utilization and user feedback.
- Refined the `install_swarm_comfy` script to skip installation when backends are already present, streamlining the setup process.
- Improved the `resolve_llm_runtime` function to prioritize live configuration over stale state notes, ensuring accurate runtime detection.
- Added tests to validate the new backend status handling and idle management logic, ensuring robustness and reliability.

src/gpu_rent/ready.py

@@ -65,11 +65,12 @@ while time.time() < deadline:
             print(f"BUSY backend={bstat} (Comfy стартует)")
         elif bstat == "running":
             print("READY backend=running")
+        elif bstat == "idle":
+            # Suspended backends still installed — first gen wakes them.
+            # Do not burn 2400s waiting for running after AllowIdle.
+            print("READY backend=idle (backends present, suspended)")
         elif bstat in ("disabled", "all_disabled"):
             print(f"BUSY backend={bstat}")
-        elif bstat == "idle":
-            # Suspended backends — not ready for generate; keep waiting.
-            print("BUSY backend=idle (бэкенды спят, ждём running)")
         elif bstat == "errored":
             print("BUSY backend=errored")
         else:
@@ -176,9 +177,9 @@ def wait_backend_idle(
         "BUSY backend=loading (Comfy стартует)": "… Comfy стартует",
         "BUSY backend=some_loading (Comfy стартует)": "… Comfy стартует (часть бэкендов)",
         "BUSY backend=empty (нужен first-install Comfy)": "… backend пуст — нужен install",
-        "BUSY backend=idle (бэкенды спят, ждём running)": "… бэкенды idle/спят",
         "BUSY backend=errored": "… backend errored — смотри journalctl -u swarmui",
         "READY backend=running": "backend ready (running)",
+        "READY backend=idle (backends present, suspended)": "backend ready (idle/suspended)",
     }
     while time.time() < deadline:
         try:
@@ -348,6 +349,9 @@ def verify_gpu_env(
         "dlbackend пуст",
         "backend=empty",
         "Install не прогоняли",
+        "available=false",
+        "cuda=none",
+        "без cuda",
     )
 
     want_swarm = bool(getattr(cfg, "enable_swarmui", True))
@@ -514,9 +518,9 @@ def verify_stack_local(
                         with urllib.request.urlopen(req, timeout=4) as resp:
                             ok = True
                             detail = f"API HTTP {getattr(resp, 'status', 200)}"
-                    except Exception:
-                        ok = True
-                        detail = f"TCP :{port} open"
+                    except Exception as exc:
+                        ok = False
+                        detail = f"HTTP/API fail (TCP open): {str(exc)[:100]}"
                 last.append(ServiceCheck(name, ok, detail, "local"))
         if last and all(c.ok for c in last):
             for c in last: