I'm wondering why "all_states" (and to also states_to_number) of the hilbert space "Spin" uses the convention

state[i] = - 2 sz[i] ?

Using "all_states" on a hilbert space of just one spin-1/2, you get the states array: [[-1],[1]]. In essence, if you implement a new type of operator (e.g. in my case PauliStrings that work with Spin instead of Qubit), you might want to check the sz[i] directly to compute the "get_conn_padded" (as is now also done for the qubit case). Wouldn't it be more straightforward to take state[i] = 2 sz[i] ?
This does not really affect much code, I think, since all operators use the 'number' indices of the states anyways. However, if you want to extend PauliStrings to Spins, it would be clearer if you change signs whenever "-1" occurs.