feat: add auto_connect flag and disconnect endpoint for computes

Computes can now be flagged auto_connect in the database. The controller
connects to flagged computes in the background at startup (reusing the
one-second delayed fire-and-forget mechanism), and enabling the flag via
PUT /computes/{id} triggers a background connection without the request
ever waiting for it. Disabling the flag does not drop an existing
connection.

Add POST /computes/{id}/disconnect to tear a compute down manually: the
notification stream and HTTP session are closed and every reconnection
path (retry backoff, notification stream reconnect, lazy connection in
http_query) is blocked until connect() is called again or the controller
restarts with auto_connect enabled. Disconnecting a compute used by an
opened project is rejected with 409 unless force=true; forcing only
drops the controller-side control channel, nodes keep running unmanaged.
The local compute cannot be disconnected.

Also make test_compute_list create its own compute (it relied on rows
from earlier tests and failed under reordering), and stop scheduling the
delayed add_compute connection during tests: the timer leaked real
connection attempts into other tests' event loops.

gns3server/api/routes/controller/computes.py

@@ -26,6 +26,7 @@ from fastapi import APIRouter, Body, Depends, Query, status
 import gns3server.db.models as models
 from gns3server import schemas
 from gns3server.controller import Controller
+from gns3server.controller.controller_error import ControllerBadRequestError, ControllerError
 from gns3server.db.repositories.computes import ComputesRepository
 from gns3server.db.repositories.rbac import RbacRepository
 from gns3server.services.computes import ComputesService
@@ -80,6 +81,42 @@ async def connect_compute(compute_id: Union[str, UUID]) -> None:
         await compute.connect(report_failed_connection=True)
 
 
+@router.post(
+    "/{compute_id}/disconnect",
+    status_code=status.HTTP_204_NO_CONTENT,
+    responses={409: {"model": schemas.ErrorMessage, "description": "Compute is used by opened projects"}},
+    # dependencies=[Depends(has_privilege("Compute.Audit"))]  # FIXME: this is a temporary workaround due to a bug in the web-ui
+)
+async def disconnect_compute(compute_id: Union[str, UUID], force: Optional[bool] = False) -> None:
+    """
+    Disconnect from a compute on the controller.
+
+    The compute is not reconnected until an explicit connection is requested,
+    or the controller restarts with auto_connect enabled for this compute.
+
+    Rejected with 409 when an opened project still uses the compute, unless
+    force is true. Forcing does not stop anything on the compute: only the
+    controller-side control channel (notification stream and HTTP session) is
+    torn down, nodes keep running unmanaged.
+
+    Required privilege: Compute.Audit
+    """
+
+    if str(compute_id) == "local":
+        raise ControllerBadRequestError("Cannot disconnect the local compute")
+    controller = Controller.instance()
+    compute = controller.get_compute(str(compute_id))
+    if not force:
+        open_projects = controller.get_compute_open_projects(compute)
+        if open_projects:
+            project_names = ", ".join(f"'{project.name}'" for project in open_projects)
+            raise ControllerError(
+                f"Compute '{compute_id}' is used by opened projects ({project_names}), "
+                "disconnect anyway with force=true"
+            )
+    await compute.disconnect()
+
+
 @router.get(
     "/{compute_id}",
     response_model=schemas.Compute,