TST: keep the Monte Carlo fixtures out of the working directory

[thc1006]

Running the unit tests leaves three zero-byte logs in the repository root, and nothing ignores them. Addresses #1189.

Pull request type

• Code maintenance (refactoring, formatting, tests)

Checklist

• Tests for the changes have been added (if needed)
• Docs have been reviewed and added / updated
• Lint (ruff check / ruff format --check) has passed locally
• All tests have passed locally
• CHANGELOG.md: nothing user-facing changes here, so no entry.

Current behavior

The two shared Monte Carlo fixtures pass a bare filename="monte_carlo_test", which resolves against whatever directory pytest was started from:

$ git status --porcelain          # nothing
$ pytest tests/unit/simulation -q
$ git status --porcelain
?? monte_carlo_test.errors.txt
?? monte_carlo_test.inputs.txt
?? monte_carlo_test.outputs.txt

test_ellipses_background_saves_file_successfully writes a fourth into the same place and removes it in a finally, which does not run if the process is killed.

New behavior

All three take tmp_path, so the logs go where pytest already cleans up after itself, and the plot test needs no cleanup clause.

A test asserts it, on filename rather than on the three file attributes: import_results repoints those at whatever it read, so they are not what the fixtures control. Putting the bare name back turns it red for both fixtures.

Breaking change

• No

Test-only.

Additional information

One thing I found while writing the assertion, and fixed with it. monte_carlo_calisto_pre_loaded also reads its example file by a path relative to the working directory. import_outputs opens that path "r+", and "w+" when it is missing, so the read target is a tracked file opened for writing and a wrong guess would create one inside the tree. It is resolved against the fixture module now. Nothing was being created today, since the file it names does exist; this is the same defect as the one above rather than a new symptom.

What this does not cover, so it is not mistaken for covered. tests/integration/simulation/test_monte_carlo.py writes four more files into the working directory, a .kml and three .png, and asserts on them by bare name. Those differ from the case above in that the tests do remove them, through _post_test_file_cleanup in a finally. The gap there is narrower: a finally does not run if the process is killed. Changing them means changing the assertions that name them, in a file this branch does not touch and whose tests are marked slow, so I have left them and am saying so rather than letting the issue look fully answered. The three log names in that cleanup list are inert after this change, and I have left those entries alone for the same reason.

While there: import_results(filename=...) hands the same name to import_outputs, import_inputs and import_errors, so after the pre-loaded fixture runs, all three attributes point at example.outputs.txt. I have left that alone, since it is the documented signature rather than anything this changes.

Verified on a50d07d5:

pytest tests/unit/simulation   333 passed, 7 skipped, 1 failed
  git status afterwards        clean
ruff check . / ruff format --check   clean

The one failure is test_ellipses_reports_an_image_path_that_does_not_exist, an imageio import this machine cannot satisfy. It is red on develop too.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.58%. Comparing base (a50d07d) to head (1bbaadf).

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #1197   +/-   ##
========================================
  Coverage    91.58%   91.58%           
========================================
  Files          132      132           
  Lines        18404    18404           
========================================
  Hits         16855    16855           
  Misses        1549     1549           

☔ View full report in Codecov by Harness.
📢 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.