coverage: invoke via pdm run for consistency with test runner

Match the convention that CLAUDE.md already documents (line 27:
"Prefix commands with `pdm run` if the venv is not active").

- `.github/workflows/coverage.yml`: replace the
  `pdm use --venv in-project && source .venv/bin/activate && ...`
  dance with `pdm run python -m coverage ...`.  Besides being
  more concise, this removes a hardcoded POSIX venv layout
  (`.venv/bin/activate`) that wouldn't work on Windows.
- `measure_coverage.sh`: likewise, prepend `pdm run ` to the bare
  `coverage` invocations so the script no longer assumes an
  already-activated venv.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: Technologicat

.github/workflows/coverage.yml

@@ -43,10 +43,8 @@ jobs:
         pdm run python -m pip install coverage
     - name: Generate coverage report
       run: |
-        pdm use --venv in-project
-        source .venv/bin/activate
-        python -m coverage run --source=. -m runtests
-        python -m coverage xml
+        pdm run python -m coverage run --source=. -m runtests
+        pdm run python -m coverage xml
     - name: Upload coverage to Codecov
       uses: codecov/codecov-action@v6
       with:

measure_coverage.sh

@@ -6,7 +6,7 @@
 # https://coverage.readthedocs.io/en/coverage-5.2.1/#quick-start
 
 echo -ne "Measuring...\n"
-coverage run --source=. -m runtests
+pdm run coverage run --source=. -m runtests
 echo -ne "Generating report...\n"
-coverage html
+pdm run coverage html
 echo -ne "Done. Open htmlcov/index.html in your browser to view.\n"