LeapHybridNLSampler not rejecting for `state_size() > maximum_number_of_states`

[JoelPasvolsky]

Description
LeapHybridNLSampler accepts models with more than maximum_number_of_states states initialized but quietly ignore the rest. Users might think the rest are being used.

To Reproduce

model = traveling_salesperson(distance_matrix=DISTANCE_MATRIX)
route, = model.iter_decisions()
with model.lock():
    model.states.resize(2)
     route.set_state(0, [0, 1, 2, 3, 4])
    route.set_state(1, [0, 1, 2, 3, 4])
results = sampler.sample(
     model,
     label='SDK Examples - TSP')

Expected it to reject: "Alex C 6 days ago Something that came up... we need to add it. And reject problems based on that num_states field in the header"

Expected behavior
Reject such problems

Environment:

• OS: WIN
• Python version: 3.11

Additional context
@alexzucca90 suggested to move to here.
I notice we actually check maximum_number_of_states property here so can easily do this in the client.

[randomir]

We could also just raise a user warning and continue with the truncated states. Seems more user friendly (but less explicit), unless you think it's really important for users to realize only one state is used (currently).