fix: correct task scheduling and simplify agent internals

Fix queue cancellation, concurrency limits and worker status updates.

Use direct Explorer imports, split the Diagnoser loop into helpers and remove unused tool registration and prompt-builder code.

Correct model caching, endpoint selection and retry handling.

tests/unit/tools/test_launch_app_tool.py

@@ -17,7 +17,7 @@ from unittest.mock import AsyncMock, MagicMock, patch
 from artemis.context import ArtemisContext
 from artemis.drivers.base import BaseDeviceDriver
 from artemis.graph.state import State
-from artemis.tools.base import ArtemisTool, ToolRegistry
+from artemis.tools.base import ArtemisTool
 from artemis.tools.mobile.launch_app import (
     LaunchApp,
     LaunchAppArgs,
@@ -49,8 +49,8 @@ def mock_state():
     return MagicMock(spec=State)
 
 
-def test_launch_app_tool_subclass_and_registry():
-    """Verify LaunchAppTool is a subclass of ArtemisTool and properly registered."""
+def test_launch_app_tool_subclass():
+    """Verify LaunchAppTool is a subclass of ArtemisTool."""
     assert issubclass(LaunchAppTool, ArtemisTool)
     assert issubclass(LaunchApp, ArtemisTool)
     assert isinstance(launch_app, ArtemisTool)
@@ -60,11 +60,6 @@ def test_launch_app_tool_subclass_and_registry():
     assert launch_app.category == "action"
     assert launch_app.args_schema == LaunchAppArgs
 
-    # Registry lookup
-    reg_tool = ToolRegistry.get("launch_app")
-    assert reg_tool is not None
-    assert isinstance(reg_tool, LaunchAppTool)
-
     # GenAI FunctionDeclaration export
     declaration = launch_app.to_genai_declaration()
     assert declaration.name == "launch_app"