Correct the one-controller claim: the drone takes over, it does not refuse

Tested on the aircraft. A second ARSDK handshake is accepted and telemetry is
redirected to it; the first session's frames stop while it still reports
connected = True. So the claim inherited from pyparrot's error text, which had
reached our error messages, tool descriptions, simulator behaviour and a test
name, was wrong in the most misleading direction: a refusal would be loud, and
this is silent.

The simulator now models the takeover by default; refusal stays available
because a client must handle a non-zero status anyway.

tests/test_live_findings.py

@@ -55,3 +55,19 @@ def test_acknowledgement_is_decided_by_data_type_not_buffer():
         frame = Frame(DataType.DATA_WITH_ACK, buffer_id, 9, b"\x00\x05\x00\x00")
         assert Frame.decode_all(frame.encode())[0].buffer_id == buffer_id
         assert 0 <= BufferId.ack_for(buffer_id) <= 255
+
+
+def test_simulator_models_takeover_not_refusal_by_default():
+    """The aircraft accepts a second controller; it does not refuse one.
+
+    Tested on the real drone 2026-10-02: session A's frames froze at 264 while
+    session B took over, and A went on reporting connected = True. The
+    simulator defaulted to refusing, which is the same unverified assumption
+    the client carried, so no test could catch the difference.
+    """
+    import dataclasses
+
+    from mcbebop.sim import FakeBebop
+
+    field = next(f for f in dataclasses.fields(FakeBebop) if f.name == "single_controller")
+    assert field.default is False, "the default must model the aircraft, not the folklore"