Idea: Refactor qiskit_to_tk

[CalMacCQ]

The qiskit_to_tk and tk_to_qiskit converters are a frequent source of bug reports and feature requests from users. I suspect for a significant number of people these converters are their first exposure to using pytket and related libraries.

The code for doing the conversion is old and has been patched together over time. Its now about 1000 lines of untidy python. I think it could make the code simpler and more maintainable if we rethink/refactor the converters perhaps splitting out some of the compilcated case handling into separate internal functions.

I think this would save us time in the long run when addressing bugs and feature requests from users. May also make it easier for new contributors to make PRs.

A couple of thoughts...

• pytket and extensions now support python 3.10, 3.11 and 3.12. This could allow us to make use of structural pattern matching (introduced in 3.10) in the circuit parsers.
• I think having circuit converters which handle higher level circuit boxes is valuable as it allows easier (and fairer) side by side benchmarking of pytket vs qiskit. Also it allows qiskit users to try out pytket with less hassle. One could of course trivially just convert a pytket circuit to a qiskit circuit by exporting to and from QASM but this translation would be very lossy and remove abstractions which can be useful for the compiler. Handling more ops also expands the range of circuits we can run on the idealised AerBackend simulator.

[CalMacCQ]

Closed this for now as I don't have any plans for imporving this code more in the immediate future beyond the changes in #382 and #391. However theres still a lot of untidy conversion code especially in the tk_to_qiskit direction.

Maybe worth coming back to this at some point.