Right now, the transpiler separates operations natively supported in PyZX from those which aren't, optimises the subset which is supported, and then splices everything back together. In particular, this is a common pattern: circuit = [A, measure, B, conditional subcircuit], where only A and B are optimised.
It might be possible to handle the measure and conditional subcircuit parts using the "ground" operation in PyZX.
Right now, the transpiler separates operations natively supported in PyZX from those which aren't, optimises the subset which is supported, and then splices everything back together. In particular, this is a common pattern: circuit = [A, measure, B, conditional subcircuit], where only A and B are optimised.
It might be possible to handle the measure and conditional subcircuit parts using the "ground" operation in PyZX.
See this paper for details: https://arxiv.org/abs/2109.06071