Fix fake backend kwargs

[nonhermitian]

Summary

Fixes #6741

Adds kwargs to backend.run() with noise model.

Details and comments

[mtreinish]

LGTM, thanks for pushing the quick fix

[jakelishman]

The test failure that occurred in 06fe680 (and a few before that) looks like it'll bite again - the bug is that in FullQiskitTestCase, the method resolution via testtools for setUp() requires that super().setUp() is called. If running locally without setting QISKIT_TEST_CAPTURE_STREAMS=1, we use BasicQiskitTestCase which resolves directly to unittest, and that doesn't care. Since the other test in this module is currently unconditionally skipped, nothing noticed that its setUp() doesn't called super().setUp() until this test was originally added to the same class.

That said, while its unconditional skip reason no longer seems to be valid (Qiskit/qiskit-aer#741 is fixed and released), the test fails by seemingly using something out-of-date, and I don't yet have the experience with the codebase to know much about that.

[mtreinish]

@jakelishman can you open an issue about this? I think that's a bunch of improvements we can/should make to the testing harness. At least for this case it's a simple to add the super() call enforcement to the BasicQiskitTestCase class too. But, the way things are split has caused issues in other places too (thinking of: Qiskit/qiskit-aer@fdf2cda ).