@Czaki suggested adding to Scientific Python instead of creating new docs for SpatialData.

A plan now would be adding a paragraph in https://spatialdata.scverse.org/en/stable/contributing.html: linking to https://lectures.scientific-python.org/advanced/debugging/index.html, some links from above and https://napari.org/stable/developers/contributing/performance/index.html.

Also Profiling section of this classes https://gitlab.mimuw.edu.pl/python-tools/debugger

From scientific python, the https://lectures.scientific-python.org/advanced/optimizing/index.html#profiling-python-code fits better.

The main question is if spatialdata should have own document about performance, or we should try to convince wider community (ex. scientific python) to have one place to store all generic information about performance monitoring in one place.

There are 3 main profilers vmprof, yappi, cProfile.

There is memray from Bloomberg to trace memory https://github.com/bloomberg/memray

The PR has the two links to the Scientific Python pages, both debugging and profiling. I think this is good enough for the SpatialData docs for now. We can consider adding a very specific profiling example with SpatialData later maybe.

For profiling in general, we can consider extending the lecture-style docs at https://lectures.scientific-python.org/advanced with your debugging examples and updating the tooling section.

We can also add an extra developer guide of profiling at https://github.com/scientific-python/cookie/tree/main/docs (although this is somewhat linked to their cookie-cutter and sp-repo-review project)