Structure splitting: Smoke test and refactor for numba

[alessandrofelder]

Description

This PR adds

• an integration smoke test and additional test data (created from a real-life dataset) to reproduce #158 and cover our structure splitting code with the tests.

• a refactor to consistently use np.ndarrays and other numba-suitable refactoring in the structure splitting code.

• [ ] Bug fix

• [x] Addition of a new feature

• [ ] Other

Why is this PR needed?

Our tests don't currently cover the case where cell detection requires structure splitting.

What does this PR do?

See description.

References

Closes #166, closes #158, closes #159

How has this PR been tested?

Run tests locally and on CI.

Does this PR require an update to the documentation?

New test and fixtures have docstrings, test data documented in docstring of new testing module.

Checklist:

• [x] The code has been tested locally
• [ ] Tests have been added to cover all new functionality (unit & integration)
• [ ] The documentation has been updated to reflect any changes
• [x] The code has been formatted with pre-commit

[alessandrofelder]

Note that CI fails in expected way, ensuring our tests now can reproduce #158.

[adamltyson]

@alessandrofelder would it be worth renaming the directory containing the new test data? The dataset label doesn't mean much to anyone other than me. Maybe something like structure_split_test, then just add a note (maybe a txt file) to highlight which dataset it's from for our records?

[alessandrofelder]

Thanks for your comments, both!

Re test data, I had that thought too, but decided against it as

• the new test data is on the same scale as existing test data
• it would be nice to get a new release over the line asap for various reasons.

But maybe it's worth reconsidering this decision? We definitely want to be using pooch soon!

[dstansby]

👍 sounds good re. test data

[dstansby]

Nice - hopefully code coverage should increase when this commit runs on the main branch!