CI: measure coverage on develop, not only on pull requests

[thc1006]

Pull request type

• ReadMe, Docs and GitHub updates

Checklist

• Tests for the changes have been added (if needed)
• Docs have been reviewed and added / updated
• Lint has passed locally
• All tests have passed locally

Current behavior

test_pytest.yaml is the only workflow that uploads to Codecov, and it runs on pull_request alone. No commit on develop ever gets a coverage report.

Codecov falls back to the nearest older report it holds and says how far behind that is. The banner on my open pull requests read:

• 2026-08-14: Report is 53 commits behind head on develop.
• 2026-08-15: Report is 61 commits behind head on develop.

It grows with every merge, and the comparison is against whatever develop looked like whenever that stale report was uploaded rather than against the base the pull request is actually built on.

.codecov.yml already names master and develop under branches: for both the project and the patch status, so the configuration expects both to be measured. It is only the trigger that never produces the reports.

New behavior

The same matrix also runs on a push to master or develop, under the same paths: filter the pull request trigger uses. Each merge leaves a report at the commit the next pull request will be compared against.

docs.yml pairs pull_request with push in the same shape, so this is the existing idiom in this repository rather than a new one.

Breaking change

• No

Additional information

Nothing in either job reads pull_request context: the only mention of the event in the file is the trigger itself, so the matrix and the CodecovUpload guard behave the same on a push. Forks still get an empty CODECOV_TOKEN and fail_ci_if_error still stays off for them; a push to develop runs in this repository, so it has the token.

Two things worth weighing:

• This roughly doubles the matrix runs, once for the pull request and once for the merge. The alternative, a single cheap leg on push, would report lower coverage than the six-leg pull request reports and read as a drop.
• A merge that touches no .py file still uploads nothing, so the banner can still show a small number. That one is accurate: coverage did not change, and Codecov walking back one commit is the right answer. What this fixes is never getting a report at all.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.53%. Comparing base (e0ff281) to head (18d4fb1).
⚠️ Report is 66 commits behind head on develop.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1162      +/-   ##
===========================================
+ Coverage    82.18%   83.53%   +1.35%     
===========================================
  Files          122      130       +8     
  Lines        16355    17086     +731     
===========================================
+ Hits         13441    14273     +832     
+ Misses        2914     2813     -101     

☔ 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.