[Breaking] Split utils into sub-modules, move typing from utils to top-level (pmv.typing)

[DanielYang59]

Feel free to move things around.

• Re-import everything from __init__ to avoid breaking
• Split tests as well
• Update the module docstring to include list of available func/...
• Sort functions alphabetically
• For test utils, hide from public interface, prefer pmv.utils.testing

[DanielYang59]

@janosh Any comment on current categorizing? 😄

[DanielYang59]

This might seem a bit overkill, but I kind of want to reserve this for more test functions like #208

[DanielYang59]

Only one task left, I would add a list of available function to the module level docstring of utils.plotting tomorrow

[DanielYang59]

Thanks! I could sleep with my eyes close now :)

[janosh]

thanks so much for your help with this! 🙏

[DanielYang59]

No problem, enjoy your day (and my sleep), thanks for reviewing and all the comments

[DanielYang59]

Looks like the fix for GitHub Pages / build is working now (one extra /), but import performance test is failing for some reason, I would have a closer look tomorrow. Also it seems I forgot to display current time (pymatviz import too slow! hard_threshold=3126.38 ms), only the expected time is shown

[janosh]

i added the expected time. it now prints

FAILED tests/performance_tests/test_import_time.py::test_import_time - Failed: pymatviz import too slow! took 3722 ms, hard_threshold=3126 ms

so at least it's not a massive regression. i'm guessing it's because i added some new modules in pymatviz.classify in #252 and #253 which now need to be parsed so perhaps the slow down is expected? still would be good to find out exactly. could also be i'm importing some module at the top-level which before was only imported at run time. maybe import sklearn.metrics as skm. if that's the case then the test did it's job perfectly :)

[DanielYang59]

i added the expected time. it now prints

Thanks for adding this, I also added one in #241.

i'm guessing it's because i added some new modules in pymatviz.classify in #252 and #253 which now need to be parsed so perhaps the slow down is expected?

I just looked into the commit history and it appears I updated the import time in a6302f2 (I'm completely confused because I cannot remember I re-ran import time benchmark in this PR at all, I originally noticed that update and thought you updated the baseline because of some improvement you made and didn't have time to ask)

I don't think this should be updated by me because I don't expect import time decrease from this PR, and by the time the import time tester was added, the monty fix had already been applied.

In a word, I'm absolutely confused why I updated the import time baseline and where that data came from (as this cannot be "accidentally" updated because the import time test would print a copiable dict instead of overwriting in place), perhaps let's revert it?

[janosh]

thanks for checking! i reverted to the old timings in bc1bc29

[DanielYang59]

Thanks for the revert, though I'm still super confused how this happened and where the data came from ...