✨ Support for Qiskit's final_layout

[burgholzer]

What's the problem this feature will solve?

When reasoning about quantum circuits there are two degrees of freedom that can always lead to problems/confusion:
An arbitrary permutation of a circuit's qubits might be applied at the very beginning of the circuit (referred to as an "initial layout") or at the end of the circuit (referred to as an "output permutation").
In almost any case, a circuit containing such permutations would still be considered equivalent to a circuit without such permutations.
This is especially relevant for verifying the equivalence of quantum circuits and the results of compilation flows in particular.

For a long time, Qiskit has supported a _layout property for its QuantumCircuit objects that captured the initial layout applied during compilation. Within the MQT we use this property to infer the initialLayout for our QuantumComputation objects.
In 0.22.0 (which is now our minimum supported version), Qiskit changed the _layout property to a proper TranspileLayout class.
0.23.0 introduced an additional final_layout component that (in some confusing fashion) captures the output permutation.
This should allow to also infer the output permutation from Qiskit.
Right now, we use the measurements in a circuit for that purpose, which is rathe prone to errors (e.g., someone not adding measurements to their circuit before compilation).

Describe the solution you'd like

Starting with 0.45.0. (which should be the next minor release), the ergonomics of the TranspileLayout class have been considerably improved (see Qiskit/qiskit#10835).
That should yield a blueprint for how to interpret the final_layout and to infer the outputPermutation from Qiskit circuits.

A potential solution would first look if the final_layout attribute is present.
If so, it would check whether the new convenient syntax (>=0.45.0) is available.
If not, it replicates the logic to parse the output permutation.
If no final layout is present, the solution should check for measurements in the circuit and infer the output permutation from there (as it currently does).

[burgholzer]

Since this issue's creation, Qiskit 1.0 has been released and we moved to depend on it.
As a consequence, implementing this should become a lot easier, as no conditional logic based on the Qiskit version is required.