♻️ Refactor Packaging and CI

[burgholzer]

Description

This PR brings a major refactoring for the overall Python packaging as well as the CI employed in the project.

Packaging

We are (finally) moving away from setuptools and towards modern scikit-build-core (https://scikit-build-core.readthedocs.io/)
This brings many advantages such as

• truly editable installs that also include the C++ part; this implies incredibly fast rebuilds that can be further improved by installing ccache.
• completely eliminates the setup.py (and the MANIFEST.in) file and all custom build code that was required previously.

As part of this change, we adopt the established best practice of a src-layout for the Python package. This implies that the Python package no longer lives in the main package directory, but is moved to src/mqt/....
This avoids several pitfalls when working with Python packages and makes the overall handling way easier.

This also means there is no direct build-time dependency on cmake and ninja anymore as this is automatically handled by scikit-build-core on demand.

As part of this rewrite, several nox sessions are also updated

CI

MQT Core has introduced reusable workflows in

• 🧑‍💻 Extensive and Reusable CI/CD Setup mqt-core#396

These have been further refined in

• 🚸 Make reusable workflows more generic mqt-core#407

Overall, this allows to

• remove almost all workflow files of this repository
• eliminate a lot of maintenance required in this repository to keep up with the latest best practices
• rely on dependabot for workflow updates once MQT Core has settled enough so that it follows proper versioning and we can use tags for the workflows here instead of @main

while at the same time:

• having automatic change detection and only triggering jobs that are really necessary; with a single pass check that can be used for branch protection
• setting up ccache for all supported workflow jobs which greatly improves compilation times
• setting up mold as a better linker under linux
• splitting jobs into as many subjobs as possible and feasible to distribute work
• automatically creating wheels for all supported platforms (including exotic platforms that need emulation) and Python versions using PyPI trusted publishing
• enabling automatic retries for flaky jobs such as MSVC builds or coverage uploads
• enabling automatic build parallelization via using Ninja wherever suitable
• running an automatic workflow using the minimal versiosn for all Python dependencies to guarantee compatibility

As a nice side effect, the new CI runs have revealed a bug in the permutation handling of the ZX checker that has subsequently been resolved.

Checklist:

• The pull request only contains commits that are related to it.
• I have added appropriate tests and documentation.
• I have made sure that all CI jobs on GitHub pass.
• The pull request introduces no new warnings and follows the project's style guidelines.

[github-actions]

Cpp-Linter Report ✔️

No problems need attention.

Have any feedback or feature suggestions? Share it here.