fix: address PR review threads (test-actions, diagnostics, docker, health, storage, plugin)

Resolves review threads:
- PRRT_kwDOO6Hdxs50R8VI (test-actions.sh: remove || echo "000" curl fallback)
- PRRT_kwDOO6Hdxs50R8VJ (test-actions.sh: JSON parse failures → FAIL not silent)
- PRRT_kwDOO6Hdxs50QdKd (diagnostics.py: sanitize raw exception text from ToolError)
- PRRT_kwDOO6Hdxs50QdKs (storage.py: unassigned uses unassignedDevices query)
- PRRT_kwDOO6Hdxs50Mwlk (docker.py: port_conflicts returns flat merged list)
- PRRT_kwDOO6Hdxs50Mwlo (docker.py: logs returns plain string not dict)
- PRRT_kwDOO6Hdxs50Mt5K (docker.py: unraid_docker logs format compatibility)
- PRRT_kwDOO6Hdxs50Mt5L (health.py: or {} null guards throughout)
- PRRT_kwDOO6Hdxs50Mt5r (docker.py: port_conflicts flat list backward compat)
- plugin.json: version synced to 0.4.4 to match pyproject.toml

Changes:
- test-actions.sh: curl exit code captured directly; JSON failures surface as FAIL
- diagnostics.py: 4 ToolError sites log exc_info=True, raise sanitized messages
- storage.py: unassigned action queries unassignedDevices instead of disks
- docker.py: logs action returns newline-joined string; port_conflicts merges
  containerPorts + lanPorts into a flat list for backward compatibility
- health.py: all nested dict lookups use `or {}` instead of `.get(k, {})` to
  handle explicit GraphQL null values

Co-authored-by: Claude <noreply@anthropic.com>

unraid_mcp/tools/health.py

@@ -135,21 +135,21 @@ async def _comprehensive_check() -> dict[str, Any]:
             return health_info
 
         # System info
-        info = data.get("info", {})
+        info = data.get("info") or {}
         if info:
             health_info["unraid_system"] = {
                 "status": "connected",
                 "url": safe_display_url(UNRAID_API_URL),
                 "machine_id": info.get("machineId"),
-                "version": (info.get("versions") or {}).get("core", {}).get("unraid"),
-                "uptime": info.get("os", {}).get("uptime"),
+                "version": ((info.get("versions") or {}).get("core") or {}).get("unraid"),
+                "uptime": (info.get("os") or {}).get("uptime"),
             }
         else:
             _escalate("degraded")
             issues.append("Unable to retrieve system info")
 
         # Array
-        array_info = data.get("array", {})
+        array_info = data.get("array") or {}
         if array_info:
             state = array_info.get("state", "unknown")
             health_info["array_status"] = {
@@ -164,9 +164,9 @@ async def _comprehensive_check() -> dict[str, Any]:
             issues.append("Unable to retrieve array status")
 
         # Notifications
-        notifications = data.get("notifications", {})
+        notifications = data.get("notifications") or {}
         if notifications and notifications.get("overview"):
-            unread = notifications["overview"].get("unread", {})
+            unread = notifications["overview"].get("unread") or {}
             alerts = unread.get("alert", 0)
             health_info["notifications"] = {
                 "unread_total": unread.get("total", 0),
@@ -178,7 +178,7 @@ async def _comprehensive_check() -> dict[str, Any]:
                 issues.append(f"{alerts} unread alert(s)")
 
         # Docker
-        docker = data.get("docker", {})
+        docker = data.get("docker") or {}
         if docker and docker.get("containers"):
             containers = docker["containers"]
             health_info["docker_services"] = {