Using `axes_sums_subject_to` on a `BinaryVariable` with size 0 results in a crash when initializing a state

```python
In [1]: from dwave.optimization import Model

In [2]: model = Model()

In [3]: b = model.binary(shape=0, axes_sums_subject_to=[(0, ">=", 1)])

In [4]: model.lock()
Out[4]:

In [5]: model.states.resize(1)

In [6]: b.state(0)
zsh: floating point exception (core dumped) ipython
```

Reproduce the crash with the Python snippet using a zero-sized BinaryVariable and axes_sums_subject_to. Trace BinaryVariable state initialization and constraint handling for the size-0 case, then check the relevant test area for a regression test; done means resizing to one state and accessing b.state(0) no longer crashes.

Written by the indexing model from the issue text.