Implement detecting and transforming index order of two body electronic integrals

[jlapeyre]

• Support AO input matrix in chemists', physicists', "intermediate", or unknown index-order convention. Convert or raise an error for unsupported input.
• Tests to cover changes.
• Updated the documentation accordingly.
• Release note.
• I have read the CONTRIBUTING document.

Summary

When constructing TwoBodyElectronicIntegrals validate the data and possibly correct it.

Closes #337

Note, we considered the following, but they will not be included here:
Support spin-orbital input matrices.
Include changing spin-orbital index convention? Maybe a subsequent PR

Details and comments

This PR will transform physicists' index-order convention to chemists'. It will also raise an error on detecting an unsupported format or an unrecognized format.

[mrossinek]

Thanks @jlapeyre, this looks like a very good start! I only skimmed through the code just now so no in detail comments yet. Just one question: I find the choice for the filename starting with _ a bit odd. Maybe we can find a more descriptive name? Since it's already inside of a utils folder I don't think we need the _utils at the end either.
What about two_body_symmetry_conversion.py? Or something along those lines?

[coveralls]

Pull Request Test Coverage Report for Build 3061967887

• 65 of 66 (98.48%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.04%) to 85.504%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
qiskit_nature/second_q/operators/tensor_ordering.py	65	66	98.48%

Totals
Change from base Build 3054069499:	0.04%
Covered Lines:	17118
Relevant Lines:	20020

---

💛 - Coveralls

[jlapeyre]

The changes made by @javabster all look great! This PR looks good to go to me.

[mrossinek]

The reason that this has not progressed further, is that I am awaiting the implementation of the PolynomialTensor (#756) and SparseLabelOp (#731) after which we can tackle #666 which is going to largely refactor the existing FermionicOp. Once this happens, we will always store the electronic integrals in physicist order, at which point these conversion utilities will come in very handy to simplify the construction of said integral containers.

I envision these utilities to be exposed very publicly to simplify the construction of PolynomialTensor instances which store electronic integrals. I am potentially even considering an electronic-dedicated subclass of this tensor class, which can handle these integral conversions as well as deal with symmetry reductions/expansions to/from 8-fold symmetry in the future.

Long story short, this will get unblocked once the above issues are resolved, at which point a good location for these utilities needs to be found in the new stack 👍

[mrossinek]

I have updated this PR and moved the symmetry tools to qiskit_nature.second_q.operators.utils where it can live for now. I might change things a bit more when working on #666 which is next up on my todo list. When doing that and updating the electronic-structure related hamiltonians and properties to use that method, I will be making use of these tools for easier construction of the PolynomialTensor objects.