✨ Option to generate mirror circuits for reliable benchmarking of devices

[burgholzer]

What's the problem this feature will solve?

When benchmarking the performance of quantum devices, it is always desirable to have a deterministic outcome of a benchmark that is efficient to verify.
While this is possible for algorithms such as Grover (has the desired entry been found), or Shor (does the result allows to factorize the integer), it is much harder in practice for all kinds of variational algorithms as these typically have a very continuous distribution of measurement outcomes that is hard to predict and verify.
Mirror circuits (https://arxiv.org/abs/2112.09853) have been proposed to address this.
They take a circuit and join it with its inverse so that any execution of the circuit is expected to yield the all-zero state $\ket{0\dots 0}$ with certainty.

Describe the solution you'd like

MQT Bench should offer the option to generate mirror circuits.
Constructing these circuits is fairly easy in code, so this shouldn't be too big of a deal.
Given how such circuits will be predominantly used for evaluating hardware, I believe it makes sense to only offer this option on the mapped level.

One should be careful that the resulting circuits are still valid and executable on the targeted device. One should also avoid to apply any further optimizations on the resulting mirror circuit as these could lead to the circuit collapsing to the empty circuit, which would defeat the whole purpose of constructing them.