Deprecating isinstance(qr[0], tuple)

[1ucian0]

This PR deprecates the isinstance for a Bit when checked for a tuple. It is some sort of hack, however, many users are checking for isinstance(qr[0], tuple) and that is not supported after #2414

qr = QuantumRegister(4)

print(isinstance(qr[0], tuple)) # True
print(isinstance(qr[0], Qubit)) # True
print(type(qr[0])) # <class 'qiskit.circuit.quantumregister.Qubit'>

By overwriting __class__, we pretend to be a tuple when isinstance is called in a context that mentions tuple.

[CLAassistant]

All committers have signed the CLA.

[mtreinish]

This looks fine to me, it's a pretty big hack to inspect the call stack and look for a reference to tuple and dynamically change the response. But, I can't think of a better way. I do wonder if there will be a performance overhead for doing this in a property on a class that will be used everywhere, but we'll find out.

Before I hit Approve I do wonder if we want to clean up the tests at all. There are a bunch of deprecation warnings being emitted because of this and other tuple vs bit things, mostly in layout.

[mtreinish]

Actually I just checked and all the warnings being emitted from the layout module are about accessing or creating the layout with tuples, not anything to do with this. We can go and clean that up later.

[hushaohan]

This has made Aqua run much slower. for example, test_amplitude_estimation used to finish within 2 minutes; now it would take more than 13 minutes.

[nonhermitian]

I can confim that this is also the reason for the slow down observed in #2489