Update to_ising and from_ising to support SparsePauliOp

[t-imamichi]

Summary

• Adds opflow option to to_ising to control the output is SparsePauliOp or Opflow's operator.
• Allows from_ising to receive quantum_info's operators.

This is part of #433.
The legacy minimum eigensolver of Terra supports only opflow (not SparsePauliOp).
We will deprecate the legacy minimum eigensover and opflow when they are deprecated in Terra.

Details and comments

[coveralls]

Pull Request Test Coverage Report for Build 4732501811

• 40 of 40 (100.0%) changed or added relevant lines in 4 files are covered.
• 1 unchanged line in 1 file lost coverage.
• Overall coverage decreased (-0.003%) to 92.083%

---

Files with Coverage Reduction	New Missed Lines	%
qiskit_optimization/algorithms/minimum_eigen_optimizer.py	1	87.69%

Totals
Change from base Build 4663084632:	-0.003%
Covered Lines:	4350
Relevant Lines:	4724

---

💛 - Coveralls

[t-imamichi]

I updated the option name from quantum_info to opflow because I thought it is easier to understand why it will be deprecated when opflow is deprecated in terra.
I also changed the code to display a warning message if opflow option is not set explicitly.

[a-matsuo]

I reviewed the code changes thoroughly and everything looks good to me!

[a-matsuo]

I reviewed the code changes thoroughly and everything looks good to me!

[woodsp-ibm]

I think directly stating the default here is easier for an end user - but we will have to remember to update this here too when the default changes. The alternative would be to say the default is determined by the to_ising function and link to that, this is the place that determines it and we have to change next time so if this were done not change would be needed here.

[t-imamichi]

Thank you for your idea. I updated it to refer to to_ising function. cc41551

[woodsp-ibm]

Just as a note given opflow is planned for deprecation in the upcoming Terra 0.24 release it would be removed in 0.26 two releases later. At that time this parameter can still exist but will be need to be ignored since there will no longer be any opflow logic in Terra. I guess we can consider updating this note when its changed over to False next time based on how things will progress.

[t-imamichi]

Yes, we need to update it depending on the opflow deprecation status of Terra in the future.

[woodsp-ibm]

Suggested change

	The ``qubit_op`` argument can currently accept Opflow operators, but it has been
	superseded by ``SparsePauliOp``.
	Opflow will be deprecated in a future release and subsequently removed after that.
	The ``qubit_op`` argument can currently accept Opflow operators (OperatorBase type), but have been
	superseded by Qiskit Terra quantum info BaseOperators such as ``SparsePauliOp``.
	Opflow operator support will be deprecated in a future release and subsequently removed after that.

Here is accepts all quantum info BaseOperator types right. Most often it will be SparsePauliOp bit its not restricted to this. I will also note that we can make these class references since we have the intersphinx with Terra.

[t-imamichi]

Thanks. I include it. cc41551

[woodsp-ibm]

I did not alter the text here as per he above suggestion. But if the suggestion seems better then perhaps this should be the same.

[t-imamichi]

Thanks. I include it. cc41551

[woodsp-ibm]

Suggested change

	Updated :func:`~.to_ising` to support ``qiskit.quantum_info.SparsePauliOp``.
	Added ``opflow`` argument to :func:`~.to_ising` to control the output is
	opflow operator (if ``True``) or ``SparsePauliOp`` (if ``False``).
	The default value of ``opflow`` argument is currently ``True``, but it will
	first be changed to ``False`` and then deprecated in future releases.
	Updated :func:`~.to_ising` to support optionally returning ``qiskit.quantum_info.SparsePauliOp``.
	This is achieved by a new ``opflow`` argument to :func:`~.to_ising` to control if the output is
	the opflow operator, as has been done in the past, (if ``True``) or ``SparsePauliOp`` (if ``False``).
	The default value of ``opflow`` argument is currently ``True``, but it will
	first be changed to ``False``. The parameter will be deprecated and removed in a future release.

[t-imamichi]

Thanks. I include it. cc41551

[woodsp-ibm]

This text is similar again to the suggestion made for the from_ising note. So again is thats ok probably we want the same/similar here too.

[t-imamichi]

Thanks. I include it. cc41551

[woodsp-ibm]

LGTM. I see the release note has a minor spelling mistake in its name support-sprase-pauli-op where its sprase instead of sparse - but since the name never shows up anywhere outside of looking at the files themselves it's of little consequence.

[t-imamichi]

Good catch. But, it is a bit complicated to rename reno. So, I will merge it without renaming it.