Why checking for only single system when providing system list in `initialize_state`

[thomasloux]

def initialize_state(  # noqa: C901, PLR0911
    system: StateLike,
    device: torch.device | None = None,
    dtype: torch.dtype | None = None,
) -> SimState:
    # TODO: create a way to pass velocities from pmg and ase

    if isinstance(system, SimState):
        return system.clone().to(device, dtype)

    if isinstance(system, list | tuple):
        if len(system) == 0:
            raise ValueError("Cannot initialize state from an empty list.")
        states = [s for s in system if isinstance(s, SimState)]
        **if len(states) == len(system) and not all(s.n_systems == 1 for s in states):
            raise ValueError(
                "When providing a list of states, to the initialize_state function, "
                "all states must have n_systems == 1. To fix this, you can split the "
                "states into individual states with the split_state function."
            )**
        if len(states) == len(system):
            return ts.concatenate_states(states)

Why does this function impose this constraint? Not sure it's a problem for anyone in practise, but this seems quite arbitrary. Actually concatenate_states handle list of state with arbitrary system sizes.

[orionarcher]

If tests pass with a zero length system I'd be fine removing it. I think this is just a hedge against edge cases.