Follow-up to TaperedQubitMapper

[mrossinek]

What should we add?

A few tasks related to the TaperedQubitMapper were decided to be left out of #1031 and instead be tackled in a follow-up. This issue gathers these remaining tasks such that we do not forget about them:

• deprecate the match_convert argument in hartree_fock_bitstring_mapped

  • This argument is no longer needed because the workflow has now changed. While this method is still being called from ElectronicStructureProblem.symmetry_sector_locator, this is now guaranteed to be provided with a QubitMapper which does not introduce any tapering (because this method is actually used to construct the TaperedQubitMapper). Given that we also change the BaseProblem.symmetry_sector_locator interface (see below), we can safely deprecate this now unneeded logic.

• Deprecate BaseProblem.symmetry_sector_locator

  • This method is no longer needed in the public API since BaseProblem.get_tapered_mapper takes care of creating a TaperedQubitMapper. Thus, it may be beneficial so simplify the public API by deprecating this method and moving its contents to an internal method which a problem may implement to support tapering.
  • I am still open to discussion on this. The reason I am a bit hesitant, is that a user building a custom Problem with a custom symmetry sector logic would need to learn about this private method. However, this I would consider more a developers job who should also be able to familiarize themselves with an internal API.

• implement _ListOrDict.wrap as suggested here

• rename all variables/arguments related to QubitConverter match the QubitMapper API

  • This is optional for now and depends on the deprecation path for the QubitConverter. But it would be nice to rename (qubit_)converter to (qubit_)mapper consistently

• start the deprecation process of QubitConverter

  • In 0.6 we will be able to pending-deprecate the QubitConverter and point the users to use the QubitMapper objects directly. In the next release (or 3 months later) we will be able to then deprecate the QubitConverter.