[refactor] improve QEOM code

[mrossinek]

What is the expected enhancement?

The implementation of the QEOM algorithm looks like it could undergo some refactoring in order to make the code more maintainable. I do not have any specific targets in mind right now, but I am opening this issue to keep this in mind for some point in the future.

Some rough ideas include:

• improve the hopping operator generation and commutativity handling:
  • right now, this is split across 5+ methods which can surely be improved to some degree
  • the code needs to handle tapering manually. This results in a very hard to follow flow of logic. This can surely be streamlined quite a bit better
  • parallel_map is being used in multiple places. We should check if we can streamline this better in order to avoid multiple iterations over the same data
• generally add more detailed/explanatory comments in the code

[mrossinek]

@Anthony-Gandon would you be interested in working on this, given your recent review of #768? Please leave a comment if you do, then I can assign this issue to you :+1:

[Anthony-Gandon]

Yes you can assign this to me. Thanks!