Remove BasicAer shot limit - QAMP

[alejomonbar]

Summary

Fixes #7634
Executing the following code gives an error because max_shots for BasicAer has a limit of 65536 shots.

from qiskit import BasicAer, QuantumCircuit

backend = BasicAer.get_backend('qasm_simulator')
circuit = QuantumCircuit(1,1)
circuit.x(0)
circuit.measure(0,0)
job = backend.run(circuit, shots=int(1e6))

Details and comments

The idea is that BasicAer does not have a limit in terms of the number of shots. Instead, the default value equals zero which works as no restriction on the number of shots.

Part of qiskit-advocate/qamp-spring-22#37

[coveralls]

Pull Request Test Coverage Report for Build 2265135965

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 84.349%

---

Totals
Change from base Build 2263072888:	0.0%
Covered Lines:	54463
Relevant Lines:	64569

---

💛 - Coveralls

[alejomonbar]

@kevinsung now this issue is passing.

[kevinsung]

@alejomonbar Please use descriptive titles for your pull requests. For example, "Remove BasicAer shot limit". Also, if you add the sentence "Fixes #7634" somewhere in the description then this PR will automatically close that issue.

[kevinsung]

@jakelishman thoughts on this solution?

[jakelishman]

Looks good to me, thanks! Please can you also add an "upgrade" release note that says that the limit has been removed?

Do you know how long the new test takes to run?

[alejomonbar]

Looks good to me, thanks! Please can you also add an "upgrade" release note that says that the limit has been removed?

Do you know how long the new test takes to run?

Thank you @jakelishman, I added the release note. About the time, I don't know if there is any command to know it, but, the circuit for the test is simple and only requires 2e6 shots, I don't think this adds much extra time for the tox test.

[jakelishman]

Two million shots is actually quite a lot, especially for a simulator implemented in pure Python. As things stood, the test took about 8 seconds. Taken alone, that's ok, but Terra's test suite has something like ten to fifteen thousand tests in it, and at that scale, 8 seconds is a long time to test on a simple toggle.

While writing the test was definitely the right thing to do, in this very particular circumstance, I think the best course of action is just to not have one. It can be hard to test for the absence of a limit efficiently, and this particular feature is something that's very unlikely to be accidentally changed - if a limit is re-instated in the future, it'll almost certainly be somebody having made a deliberate technical choice.

Thanks for the contribution!