Separate test utils from tests

[flying-sheep]

This is basically the minimum amount of changes to separate things out and fix some problems with the test setup, plus unification of how we handle optional dependencies in tests.

Fixes #2225

[codecov]

Codecov Report

Merging #2235 (fcb0a06) into master (06802b4) will increase coverage by 0.24%.
The diff coverage is 85.84%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2235      +/-   ##
==========================================
+ Coverage   71.89%   72.14%   +0.24%     
==========================================
  Files          98      104       +6     
  Lines       11518    11678     +160     
==========================================
+ Hits         8281     8425     +144     
- Misses       3237     3253      +16     

Impacted Files	Coverage Δ
scanpy/testing/_pytest/__init__.py	60.86% <60.86%> (ø)
scanpy/testing/_pytest/marks.py	84.61% <84.61%> (ø)
scanpy/testing/_helpers/data.py	88.57% <88.57%> (ø)
scanpy/plotting/_tools/__init__.py	76.77% <100.00%> (ø)
scanpy/plotting/_tools/scatterplots.py	87.20% <100.00%> (-1.22%)	⬇️
scanpy/testing/_helpers/__init__.py	100.00% <100.00%> (ø)
scanpy/testing/_pytest/fixtures/__init__.py	91.66% <100.00%> (ø)
scanpy/testing/_pytest/fixtures/data.py	100.00% <100.00%> (ø)

... and 5 files with indirect coverage changes

[ivirshup]

I've written a bit more about this in the issue thread, but I would be up for a much more narrowly scoped PR. But not something this broad.

Moving utilities out of the test suite, I'm up for. Everything else, not so much.

[flying-sheep]

@ivirshup The test failures are a bug exposed by the fixture refactoring. The tests were relying on adata['uns']['pos'] being left over from a previous test run. Can you help me fix it?

[ivirshup]

@flying-sheep, I see you're doing some updates here.

Are you doing anything to narrow the scope of the PR as requested last go-around?

[flying-sheep]

I answered that back then already.

• Separate test utils from tests #2235 (comment)
• Separate test utils from tests #2235 (comment)

This PR has exactly the scope necessary to separate test utils and tests.

The only thing that I can think of to add to those answers is that the repeated code for all the data fixtures is necessary to make editors understand them. A more dynamic way to make all those fixtures breaks ctrl/cmd-clicking fixtures.

[flying-sheep]

OK, everything done as we agreed on yesterday:

• only parametrized datasets are fixtures
• the skip mark is now a single function.

[lumberbot-app]

Owee, I'm MrMeeseeks, Look at me.

There seem to be a conflict, please backport manually. Here are approximate instructions:

1. Checkout backport branch and update it.

git checkout 1.9.x
git pull

2. Cherry pick the first parent branch of the this PR on top of the older branch:

git cherry-pick -x -m1 e5d41d4aa58a925f0fa5cfcf580cb975167a71c9

3. You will likely have some merge/cherry-pick conflict here, fix them and commit:

git commit -am 'Backport PR #2235: Separate test utils from tests'

4. Push to a named branch:

git push YOURFORK 1.9.x:auto-backport-of-pr-2235-on-1.9.x

5. Create a PR against branch 1.9.x, I would have named this PR:

"Backport PR #2235 on branch 1.9.x (Separate test utils from tests)"

And apply the correct labels and milestones.

Congratulations — you did some good work! Hopefully your backport PR will be tested by the continuous integration and merged soon!

Remember to remove the Still Needs Manual Backport label once the PR gets merged.

If these instructions are inaccurate, feel free to suggest an improvement.