🧪 CI: upload coverage on pushes to master too

[chrisjsewell]

Every Codecov upload step is guarded by github.event.pull_request.head.repo.full_name == github.repository, a guard against fork pull requests (#1229, August 2024) that has no value on a push event. So no push to master has uploaded a report since then; measured on the CI run for cc0ffe0, all five "upload to Codecov" steps are skipped.

Codecov compares a pull request against the newest ancestor that has a report. That ancestor is now d68d10d, 35 commits back and from before the sphinx-test-reports split (#2009), which moved statements out of the reports flag, so codecov/project/reports reads as a coverage drop (−0.74%) on every pull request, including ones that never touch the package (#2020 is the current example). A looser threshold in codecov.yml would hide this one delta while the base kept drifting; the cause is the missing upload.

What changes

• The fork guard applies to pull-request events only: (github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository). A push to master uploads under the repository's own secrets; the dependabot and repository checks are unchanged.
• fail_ci_if_error is limited to pull requests, so a Codecov outage cannot turn master red.
• The two CODECOV_TOKEN secret descriptions say when the steps read it.
• The reason is recorded in the step comment, with the history (Deactivates codecov service usage for forks #1229 wrote the guard, ‼️ Split sphinx-test-reports: the Sphinx-free core becomes ub-test-reports 1.0.0 #2009 copied it onto the ub-test-reports step).

What to expect

The first push to master after this merges uploads a fresh report; pull requests opened or rebased after that compare against it. Pull requests whose base is older, #2020 included, keep their current reports status until they rebase or merge; it is not a finding about them.

Split out of #2020 at review's request; that branch reverts its copy of the change.

---

Generated by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.87%. Comparing base (d68d10d) to head (46eb8f5).
⚠️ Report is 36 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2023      +/-   ##
==========================================
+ Coverage   91.69%   91.87%   +0.17%     
==========================================
  Files         129      129              
  Lines       18155    18084      -71     
==========================================
- Hits        16648    16614      -34     
+ Misses       1507     1470      -37     

Flag	Coverage Δ
codelinks	93.74% <ø> (+0.32%)	⬆️
mounts	94.26% <ø> (+0.24%)	⬆️
pytests	91.48% <ø> (+0.07%)	⬆️
reports	88.01% <ø> (-0.74%)	⬇️
ub-test-reports	90.16% <ø> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.