Fix from_others issue with SQLgraph.

[yfukai]

Graphs generated by RegionPropsNodes could not be converted from RustWorkXGraph to SQLGraph. This PR intends to solve this issue. This is a quick fix, and later it may make sense to refactor the overall logic.

I also updated curry to partial since I believe the later is sufficient here.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 81.25000% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 88.17%. Comparing base (cb4770a) to head (5939c92).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/tracksdata/nodes/_regionprops.py	20.00%	2 Missing and 2 partials ⚠️
src/tracksdata/utils/_dtypes.py	85.71%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #182      +/-   ##
==========================================
- Coverage   88.22%   88.17%   -0.05%     
==========================================
  Files          51       51              
  Lines        3593     3612      +19     
  Branches      617      625       +8     
==========================================
+ Hits         3170     3185      +15     
- Misses        254      256       +2     
- Partials      169      171       +2     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[JoOkuma]

@yfukai, I might have introduced this bug in #170.

I'll carefully review the pickling behavior from your PR

Ideally, and I think it's still the case, when an array has a fixed length, it should return a polars series of columns of array or list. I will add some tests for this.

[JoOkuma]

Hey @yfukai, I was able to simplify your changes due to #170, but I broke the tests when merging with main.
My changes are at yfukai#7. Can you see if you agree with it?

[yfukai]

Maybe this should be inherited from the other graph? Or we can handle this with a new PR.

[JoOkuma]

@yfukai, unfortunately, we don't store the default value at the graph level. This is something I want to add when reworking the add_node_attr_key to have something better.
Do you see another way of getting it from the other graph?

[yfukai]

I merged yfukai#7 and started to work on that! I left a comment about the default value.

[JoOkuma]

@yfukai, thanks. I left a reply to your comment. If you can inherit the existing information, that would be great. I'm going to rework the add_{node|edge}_attr_key soon, so this will soon change.

Let's remove the centroid prop from the failing test.

I'm trying to fix the benchmark at #190 , without much success.

[yfukai]

Hi @JoOkuma, Thanks!
Can you check if yfukai#8 makes sense? I think current impl fails with non-array and non-None pickable objects.

This is something I want to add when reworking the add_node_attr_key to have something better.

This sounds great! I don't have an idea other than saving a dict of attr_key -> default_value mapping in graphs...

[yfukai]

We can work on regionprops default value a bit more (since some props are array-typed) but I created another issue (#191) for that.

[yfukai]

Hi @JoOkuma thank you for your comment! For me this PR is fine to be merged!

[JoOkuma]

Hi @yfukai, I'm merging this PR. The benchmark error is unrelated; I'm trying to fix it on #190