Implement FermionicOp.from_polynomial_tensor

[mrossinek]

What should we add?

#665 proposes the addition of a PolynomialTensor class to handle coefficient storage.

This issue tracks the implementation of the method which can construct a FermionicOp from those coefficients.
Below, I summarize the proposed structure of the PolynomialTensor.data field:

Currently, coefficients that will later become a FermionicOp are stored in most cases in the ElectronicIntegrals. Those are tailored to the electronic structure problems allowing some memory reduction in scenarios of identical coefficients for spin-up and -down orbitals, rendering them not generally applicable.
Instead, I propose a structure of data as follows:

{
  "+-": [
    [1, 2],
    [3, 4]
  ],
  "++--": [
    [[
      [1, 2],
      [3, 4]
    ], [
      [5, 6],
      [7, 8]
    ]], [[
      [9, 10],
      [11, 12]
    ], [
      [13, 14],
      [15, 16]
    ]]
  ]
}

Here +- stores a one-body term where the first axis of the stored matrix will be mapped to + operators, and the second axis to -. I.e. this will result in a FermionicOp:

1.0 * "+_0 -_0" + 2.0 * "+_0 -_1" + 3.0 * "+_1 -_0" + 4.0 * "+_1 -_1"

Accordingly, ++-- indicates a two-body term, and, even more importantly, unambiguously stores the two-body terms in the setting of electronic structure problems, in the physicist notation:

$$ g_{pqrs} a^{\dagger}_p a^{\dagger}_q a_r a_s $$

For brevity, I will not expand the 16 coefficients into the operator but will only show a few examples:

• data["++--"][0, 0, 0, 0] -> 1.0 * "+_0 +_0 -_0 -_0"
• data["++--"][0, 1, 0, 1] -> 6.0 * "+_0 +_1 -_0 -_1"
• data["++--"][1, 0, 0, 0] -> 9.0 * "+_1 +_0 -_0 -_0"

For the specific case of electronic structure problems, we can implement a specialized variant of the PolynomialTensor which will allow simplified construction from chemistry-notation two body integrals (using the utilities developed in #520), can implement memory-reduction methods to exploit spin symmetries, and (in the future) could also exploit further symmetry information in chemical systems (like 8-fold symmetry of AO ERIs, for example).