Update built-in pass managers (followup of #11448)

[nkanazawa1989]

Summary

This PR is a followup of #11448 to keep logical change minimum. This PR overhauls built-in pass managers. PassManager.append pattern is almost removed from them since this is no longer necessary.

Details and comments

PassManager can be instantiated with predefined pass sequence even though they include conditional or do-while controllers as follows

pm = PassManager(
    ConditionalController(
        DoWhileController(
            [my_pass1, my_pass2, my_pass3], 
            do_while=callable2
        ), 
        condition=callable1
    )
)

The append pattern may be necessary when the pipeline changes with some condition (conditional branching):

if some_condition:
  pm.append(some_passes)
else:
  pm.append(some_other_passes)

In principle this can be also updated by introducing something like IfElseController. Then PassManager.append can be also dropped like flow controllers. However this is not the scope of this PR.

In addition, some inefficiency is removed with this PR. In the current code, following pattern is often used

pm_stage += some_other_pm

this implicitly builds new instance instead of mutating the original instance, thus it always comes with extra memory overhead (the original instance is discarded by GC though).

[qiskit-bot]

One or more of the the following people are requested to review this:

• @Qiskit/terra-core

[nkanazawa1989]

Blocked by #11448

[coveralls]

Pull Request Test Coverage Report for Build 7580069842

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.001%) to 89.495%

---

Totals
Change from base Build 7576041329:	-0.001%
Covered Lines:	59411
Relevant Lines:	66385

---

💛 - Coveralls

[jakelishman]

Sorry I didn't get to this sooner. Since it's just an internal code refactor, it's eligible for 1.1 and I'll defer it til then.

[ElePT]

Thanks a lot for opening this PR @nkanazawa1989, unfortunately, I believe that the code base has diverged quite a bit from the original state that was considered in this PR. I will close it for backlog managing purposes.