MULTILAYER's main functionality (the one we would want to benchmark) is GUI-based. Unfortunately, the GUI stuff is not just a thin shim that collects arguments and passes them to some class that we could easily trigger without the GUI. And all of the code is in a 3.6k line single module that will be hard to patch - while making sure we are retaining the intended functionality (at least I cannot do that with my zero knowledge of spatial transcriptomics). My recommendation is to either skip this method or ask the dev to create a CLI- or proper library-type entry point that we can hook in.

In any case, I will create a PR for the environment file I created (which was also a pain) - in case anyone feels up for a challenge 💪