[Feature] Adding apply to SummedPauliOp

[jamesETsmith]

General

This PR add SummedPauilOp.apply for the operation: $\sum_k \sum_j h_{jk} \mathcal{P}_j \ket{\psi_t}$ and has some small performance tweaks for SummedPauliOp.apply_weighted.

Details

• Implements SummedPauilOp.apply
• Small bug fix for apply_weighted and expectation_value where the ExecutionPolicy wasn't being checked properly and it was always running serially

Update on SummedPauliOp.square

I've implemented the square function to create a SummedPauliOp from sparse utilities that's equivalent to (A_k)^2. We do not save the "map" t_aij like we'd want to if we call this function immediately, but introducing that capability in the future should be straightforward if we want to.

In terms of performance,

• $\big( A_k \big)^2 \ket{\psi}$ is slower than $A_k \big( A_k \ket{\psi} \big)$ in all situations I looked at
• $A_k$ --> dense --> $A_k^2$ is faster than $A_k$ --> $A_k^2$ --> dense for smaller numbers of qubits, but around 6-8 qubits, the latter becomes much faster

Remaining TODOs

• Implement to_tensor
• Implement optimized version to apply a SummedPauliOp squared to a set of states
• Add C++ and Python docstrings for SummedPauliOp methods
• C++ tests for new methods
• Python tests for new methods

---

• To see the specific tasks where the Asana app for GitHub is being used, see below:
  • https://app.asana.com/0/0/1208613694344929

[stand-by]

LGTM 👍
Thank you James