check.py for running avocado coverage.py

The avocado was using `selftests/run` script for running coverage tests.
This script uses python unittests module instead of avocado, which has
few disadvantages. The main is that running the `selftests/run` differs
from running `check.py`, therefore the coverage report hasn't
represented the actual state.

After the latest changes in ea0e23d  and 4100cc2 it is possible to run
avocado's main selftests entry point `check.py` with the coverage.py.
With such change, the running coverage diagnose will use the same
process as our selftests, therefore the report will be more precise.

Reference: #4992
Signed-off-by: Jan Richter <[email protected]>

.coveragerc

@@ -0,0 +1,2 @@
+[run]
+source = avocado/, optional_plugins/

.github/workflows/ci.yml

@@ -132,7 +132,9 @@ jobs:
       - name: Run pre script
         run: ./cc-test-reporter before-build
       - name: Run script
-        run: make develop && ./selftests/run_coverage
+        run: |
+         python setup.py develop --user
+         ./selftests/run_coverage
       - name: Run post script
         run: ./cc-test-reporter after-build
       - run: echo "🥑 This job's status is ${{ job.status }}."

selftests/run_coverage

@@ -15,8 +15,8 @@ echo "Using coverage utility: $COVERAGE"
 
 $COVERAGE erase
 rm -f .coverage.*
-RUNNING_COVERAGE=1 AVOCADO_CHECK_LEVEL=1 UNITTEST_AVOCADO_CMD="$COVERAGE run -p --include 'avocado/*,optional_plugins/*' $PYTHON -m avocado" $COVERAGE run -p --include "avocado/*,optional_plugins/*" ./selftests/run
-$COVERAGE combine .coverage*
+$COVERAGE run selftests/check.py --skip=static-checks
+$COVERAGE combine
 echo
 $COVERAGE report -m --include "avocado/core/*"
 echo