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.

tests/test_tunnel_watch.py

@@ -27,30 +27,34 @@ def test_decide_exit_missing():
     assert d.kind == "exit"
 
 
-def test_tunnel_forwards_swarm_only(monkeypatch):
+def test_decide_soft_fail_keeps_tunnel():
+    d = decide_watch("SOFT_FAIL", tunnel_alive=True)
+    assert d.kind == "ok"
+    assert "soft-fail" in d.detail
+
+
+def test_tunnel_forwards_swarm_only():
     class Cfg:
         swarmui_local_port = 17801
         llm_runtime = "none"
         ollama_local_port = 17811
+        enable_swarmui = True
 
-    monkeypatch.setattr("gpu_rent.tunnel.load_state", lambda: type("S", (), {"notes": {}})())
     assert tunnel_forwards(Cfg()) == [(17801, 7801)]
 
 
-def test_tunnel_forwards_prefers_cfg_over_stale_notes(monkeypatch):
+def test_tunnel_forwards_prefers_cfg_over_stale_notes():
+    """tunnel_forwards uses cfg only — notes must not add Ollama."""
     class Cfg:
         swarmui_local_port = 17801
         llm_runtime = "none"
         ollama_local_port = 17811
+        enable_swarmui = True
 
-    monkeypatch.setattr(
-        "gpu_rent.tunnel.load_state",
-        lambda: type("S", (), {"notes": {"llm_runtime": "ollama"}})(),
-    )
     assert tunnel_forwards(Cfg()) == [(17801, 7801)]
 
 
-def test_resolve_llm_notes_only_when_cfg_none(monkeypatch):
+def test_resolve_llm_uses_cfg_only(monkeypatch):
     from gpu_rent.access_card import resolve_llm_runtime
 
     class Cfg:
@@ -60,13 +64,10 @@ def test_resolve_llm_notes_only_when_cfg_none(monkeypatch):
         "gpu_rent.access_card.load_state",
         lambda: type("S", (), {"notes": {"llm_runtime": "ollama"}})(),
     )
-    assert resolve_llm_runtime(Cfg()) == "ollama"
+    # Stale notes must not override live cfg=none
+    assert resolve_llm_runtime(Cfg()) == "none"
 
     class Cfg2:
         llm_runtime = "ollama"
 
-    monkeypatch.setattr(
-        "gpu_rent.access_card.load_state",
-        lambda: type("S", (), {"notes": {"llm_runtime": "none"}})(),
-    )
     assert resolve_llm_runtime(Cfg2()) == "ollama"