Support Qiskit 1.x

[speller26]

Summary

Removes qiskit<1.0 restriction and updates tests to use 1.x-compatible components.

Details and comments

[jcjaskula-aws]

LGTM at first pass. While I have a more thorough look in the coming days, can you ensure me that the notebooks run well top to bottom?

[speller26]

Tests are failing due to incorrect implementation of MS gate in qiskit-ionq: qiskit-community/qiskit-ionq#162. Will be fixed by qiskit-community/qiskit-ionq#174.

[jcjaskula-aws]

Minor comments. I'll be fine if you decide to stick with your current implementation.

[jcjaskula-aws]

Should we write somewhere if there is anything we expect to change in our code when we switch to 0.5.0?

[speller26]

There's also a TODO in the relevant code:

qiskit-braket-provider/tests/providers/test_adapter.py

Lines 170 to 175 in 79f229a

	# TODO: use compare_braket_qiskit_unitaries once
	# MS gate implementation in qiskit-ionq has been corrected
	np.allclose(
	to_braket(qiskit_circuit).to_unitary(),
	Operator(qiskit_circuit).to_matrix(),
	)

[jcjaskula-aws]

Nit: if it is useful somewhere else, we could have compare_braket_qiskit_unitaries(braket_circuit, qiskit_circuit) and call compare_braket_qiskit_unitaries(to_braket(qiskit_circuit), qiskit_circuit)) below.

[jcjaskula-aws]

We could also consider calling it compare_unitaries.

[speller26]

That could be useful, but the point of this method is to check whether the converted circuit has the correct matrix; I'll rename the function accordingly.