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/agents/test_explorer_perception_regressions.py

@@ -37,7 +37,7 @@ def _perception_host() -> PerceptionToolsMixin:
 async def test_get_ocr_list_reads_the_screen_index_without_storage() -> None:
     host = _perception_host()
 
-    with patch("artemis.agents.explorer.explorer.StorageManager") as storage_cls:
+    with patch("artemis.agents.explorer.run_setup.StorageManager") as storage_cls:
         result = await host.exec_get_ocr_list()
 
     assert result == {