q_state.bsz is not fix!

[s110m]

Consider the following cell in the sec1_basic.ipynb file:

# q_state = tq.QuantumState(n_wires=2)
q_state = tq.QuantumDevice(n_wires=2, bsz=1, device="cpu")
print(q_state)

# q_state.set_states(torch.tensor([[0, 0, 1, 0], [0, 1, 0, 0]]))
q_state.set_states(torch.tensor([[0, 0, 1, 0], [0, 1, 0, 0], [-5, 2, 5, 7]], dtype=torch.complex64))
print(q_state)

q_state.x(wires=0)
print(q_state)

bitstring = tq.measure(q_state, n_shots=1024, draw_id=2)
print(bitstring)

Although q_state.bsz is initially set to 1, it is still possible to pass more than one state to

q_state.set_states(torch.tensor([[0, 0, 1, 0], [0, 1, 0, 0], [-5, 2, 5, 7]], dtype=torch.complex64))

Is this behavior intentional (by design)?
It seems more consistent if the number of states matched bsz. Am I correct?

Also, we have the state [-5, 2, 5, 7] which is not normalized. Shouldn't we normalize this state when we want to consider this as a state?

Thank you.