Improve error messaging when observable is outside the range of qubits defined

Required prerequisites

• Search the issue tracker to check if your feature has already been mentioned or rejected in other issues.

Describe the feature

The code below throws up an error message as expected as we are running a 2 qubit circuit but have spin.x(2) in the hamiltonian.

However, the error message is not descriptive of the issue and can cause some headache when debugging.

import cudaq
from cudaq import spin
import numpy as np 

cudaq.set_target("nvidia")

qubit_count = 2

kernel, parameters = cudaq.make_kernel(list)

qubits = kernel.qalloc(qubit_count)

for i in range(qubit_count): 
    kernel.rx(parameters[i], qubits[i])
    
hamiltonian = spin.z(0) + spin.x(1) + spin.x(2)

parameters = np.random.rand(qubit_count)

expectation_value = cudaq.observe(kernel, hamiltonian, parameters).expectation()

expectation_value

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
Cell In[1], line 20
     16 hamiltonian = spin.z(0) + spin.x(1) + spin.x(2)
     18 parameters = np.random.rand(qubit_count)
---> 20 expectation_value = cudaq.observe(kernel, hamiltonian, parameters).expectation()
     22 expectation_value

RuntimeError: [custatevec] %invalid value in applyGateMatrix (line 172)

Thanks team

Run the Python reproducer shown in the issue, starting at cudaq.observe with the two-qubit kernel and the out-of-range spin.x(2) term. Trace where the custatevec error is surfaced and make the resulting message identify that the observable references a qubit outside the circuit range; add coverage if the existing test location is found.