Added GenIsing.py

[prasanthcakewalk]

This PR extends QTensor to perform QAOA for general Ising Hamiltonians, which can contain constant, linear, and quadratic terms.

The Ising Hamiltonian is to be encoded as a networkx graph.
For a Hamiltonian H = \sum_{i ≠ j} (A_ij/2) s_i s_j + \sum_i B_i s_i + C

1. The nodes of the graph correspond to the variables s_i = -1 or 1.
2. A_ij is given by the 'weight' attribute of the edge between i and j. If an edge is absent, the corresponding A_ij is assumed to be 0.
3. B_i is given by the 'weight' attribute of the node i. If the 'weight' attribute is absent for a node, the corresponding B_i is assumed to be 0.
4. C is specified by a graph attribute called 'offset'.

Also included is a utility to convent a docplex qubo model (for an objective function defined on binary variables x_i = 0 or 1) into a networkx graph as-per the specifications above.

[danlkv]

Hi @prasanthcakewalk, thanks for the contribution! I added a qubo_simulations branch, which is a dev with merged changes from master.

TODO, as discussed:

1. Add tests, compare to qiskit or a well-known result
2. Better handling of single-node energy

Problems to think about:

1. pynauty fails to install. @prasanthcakewalk Could you provide more info on the issue? @rsln-s did you face any problems with pynauty python package? Maybe we should consider removing from setup.py dependencies and make it an optional dependency.
2. A lot of subclasses to support different builders and composers. Consider changing the architecture.

[danlkv]

Closing as outdated, see #34