Implement idle-killer enhancements and swarm management improvements

- Marked critical bugs as resolved in the review documentation, including changes to the `arm_idle_killer` function to raise errors on credential creation failures and ensure proper file permissions for JSON credentials.
- Introduced a new `_try_arm_idle_killer` function in `provision.py` to manage idle-killer state more effectively, ensuring it arms correctly during provisioning.
- Updated the `swarm_busy` function in `remote/idle_killer.py` to allow idle state after a specified duration of Swarm unavailability, preventing unnecessary billing.
- Enhanced performance tuning logic in `tune_swarm_perf.py` to ensure proper handling of pip installation success before applying extra arguments.
- Added tests to validate the new idle-killer behavior and swarm management logic, ensuring robustness in handling idle states and error conditions.

src/gpu_rent/remote/tune_swarm_perf.py

@@ -106,17 +106,18 @@ def patch_backends_extra_args(extra: str) -> bool:
     return True
 
 
-def pip_install_sage(pip: Path) -> None:
+def pip_install_sage(pip: Path) -> bool:
     print(f"pip install triton sageattention via {pip}")
     env = dict(os.environ)
     env["PIP_DISABLE_PIP_VERSION_CHECK"] = "1"
-    # Best-effort: do not fail whole tune if wheels missing for this torch.
     cmd = [str(pip), "install", "-U", "triton", "sageattention"]
     try:
         subprocess.check_call(cmd, env=env)
         print("triton + sageattention installed")
+        return True
     except subprocess.CalledProcessError as exc:
-        print(f"WARN: pip install failed ({exc}) — ExtraArgs may no-op until fixed")
+        print(f"WARN: pip install failed ({exc}) — ExtraArgs NOT applied")
+        return False
 
 
 def main() -> int:
@@ -134,24 +135,25 @@ def main() -> int:
             return 0
 
     print(f"perf tune: {plan['name']} tier={plan['tier']} sage={plan['use_sage']}")
-    pip_ok = False
+    pip_ok = not plan["use_sage"]
     restarted_needed = False
     if plan["use_sage"]:
         pip = find_pip()
         if pip:
-            pip_install_sage(pip)
-            pip_ok = True
+            pip_ok = pip_install_sage(pip)
         else:
             print("Comfy venv pip not found yet — will retry next up")
-        if patch_backends_extra_args(plan["extra_args"]):
+            pip_ok = False
+        # Only patch ExtraArgs when wheels installed — otherwise Comfy may break.
+        if pip_ok and patch_backends_extra_args(plan["extra_args"]):
             restarted_needed = True
 
     marker = {
         "uuid": plan["uuid"],
         "name": plan["name"],
         "tier": plan["tier"],
-        "extra_args": plan["extra_args"],
-        "pip_ok": pip_ok or not plan["use_sage"],
+        "extra_args": plan["extra_args"] if pip_ok else "",
+        "pip_ok": pip_ok,
         "restart_needed": restarted_needed,
     }
     MARKER.write_text(json.dumps(marker, indent=2) + "\n", encoding="utf-8")