feat: remove save logic and use pymmcore-plus writer from MDAWidget

[fdrgsp]

This PR only removes the saving logic that we had here in napari-micromanager and uses the pymmcore_plus writers implemented in the MDAWidget (pymmcore-plus/pymmcore-widgets#279).

see also #316

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 78.49%. Comparing base (e321203) to head (ec2a4e4).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #315      +/-   ##
==========================================
- Coverage   79.22%   78.49%   -0.73%     
==========================================
  Files          15       13       -2     
  Lines         693      623      -70     
==========================================
- Hits          549      489      -60     
+ Misses        144      134      -10     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[tlambert03]

never import private stuff from other packages, even if it's your package. I know it's annoying. but it just always comes back to bite you. I'll update this one

[tlambert03]

ok, i've had a look over and it looks good!

I did add back the test_saving_mda test (we need to have at least one test to ensure we're saving things properly, even if it's not handled here). But in doing so, I hit one error in the expected data shape. You should see the error if you run:

pytest -k 'test_saving_mda and no_splitC'

        assert dest.exists()
>       assert data_shape == expected_shape
E       assert [3, 7, 2, 500, 512] == [3, 2, 7, 500, 512]
E
E         At index 1 diff: 7 != 2
E         Use -v to get more diff

any thoughts as to why the data is transposed?

[tlambert03]

figured it out. the mda_widget appears to be changing the axis_order of our requested mda. looking into it

[tlambert03]

fixes by pymmcore-plus/pymmcore-widgets#286

[tlambert03]

looks like we're now hitting this bug: pymmcore-plus/pymmcore-plus#334

so there really is something to that one. Will see if I can recreate in pymmcore-plus and fix

[tlambert03]

bug is fixed upstream, all passing now, will merge