Adds documentation for the default_filter_criterion

[Anthony-Gandon]

Fix #643

Summary

Adds a note in the docstring of the default_filter_criterion() method to explicit that it is assuming singlet spin configuration.
An exemple of how a custom filter could be given for other configurations is also provided (it is only an attempt because I don't yet know how to ensure that the AngularMomentum property exists)

Details and comments

def custom_filter_criterion(problem):
      def filter_criterion(problem, eigenstate, eigenvalue, aux_values):
          try:
              num_particles_aux = aux_values["ParticleNumber"][0]
              total_angular_momentum_aux = aux_values["AngularMomentum"][0]
          except TypeError:
              num_particles_aux = aux_values[0][0]
              total_angular_momentum_aux = aux_values[1][0]
          particle_number = cast(
              ParticleNumber,
              problem.grouped_property_transformed.get_property(ParticleNumber),
          )
          total_angular_momentum = cast(
              AngularMomentum,
              problem.grouped_property_transformed.get_property(AngularMomentum),
          )
          return np.isclose(
              total_angular_momentum.spin, total_angular_momentum_aux
          ) and np.isclose(
              particle_number.num_alpha + particle_number.num_beta, num_particles_aux
          )

      return partial(filter_criterion, problem)

At the moment and with a simple setup for the problem, I'm getting None when trying to access the AngularMomentum property.

qubit_op = QubitConverter(mapper=JordanWignerMapper(),
                                               two_qubit_reduction=False,
                                               z2symmetry_reduction=None)
driver = PySCFDriver(atom="H .0 .0 .0; H .0 .0 0.75",
                     unit=UnitsType.ANGSTROM,
                     charge=0,
                     spin=0,
                     basis="sto3g")
esp = ElectronicStructureProblem(driver)
new_numpy_solver = NumPyEigensolverFactory(filter_criterion=custom_filter_criterion(esp))
new_numpy_excited_states_calculation = ExcitedStatesEigensolver(qubit_op, new_numpy_solver)

How could I setup a working test for this?

[coveralls]

Pull Request Test Coverage Report for Build 2801582482

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 85.666%

---

Totals
Change from base Build 2799719849:	0.0%
Covered Lines:	17081
Relevant Lines:	19939

---

💛 - Coveralls

[woodsp-ibm]

We will have to ensure this gets this into the relocated classes too from the refactor of #722

[mrossinek]

As previously mentioned by Steve, we have restructured the code base, meaning that this change now needs to be applied to the class located at qiskit_nature/second_q/problems and needs to leave the old code unchanged. A matching test location exists, too.

Other than that, this PR looks good to me 👍

[mrossinek]

Thanks, @Anthony-Gandon!

[mrossinek]

Other than this, this LGTM 👍

[mrossinek]

Thanks, Anthony!