fix: fresh install archives stale state of a previous installation; a fatal Agent loop error stops the service instead of leaving a dead 'Running' one; 1.2.3

Found by uninstall + fresh install on the clean Windows Sandbox: the old agent_state.sqlite identity conflicted with the new enrollment and the service sat idle.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

installer/install_common.py

@@ -333,3 +333,19 @@ def audit_event(agent_token: str, hub: str, event_type: str, message: str) -> No
         response.raise_for_status()
     except Exception:  # noqa: BLE001 — best effort only
         return
+
+
+def archive_stale_state(install: Path) -> list[str]:
+    """A NEW enrollment (fresh install with a code) means a NEW identity. A state database left behind by a previous installation (uninstall keeps local data) still carries
+    the old installation_id, which the Agent refuses to mix with the new credential: rename it aside (never delete) so the new installation starts clean."""
+    moved = []
+    stamp = time.strftime("%Y%m%d-%H%M%S")
+    for name in ("agent_state.sqlite", "agent_state.sqlite-wal", "agent_state.sqlite-shm", "agent_state.sqlite-journal"):
+        path = Path(install) / name
+        if path.exists():
+            target = path.with_name(f"{name}.previous-{stamp}")
+            path.rename(target)
+            moved.append(target.name)
+    if moved:
+        ilog(f"archived state from a previous installation: {', '.join(moved)}")
+    return moved