chore: improve flaky rerun step in CI (pytest crash, appended coverage)

Makefile

@@ -235,8 +235,7 @@ pytest_internal_parallel:
 # --global-coverage-infos-json=global-coverage-infos.json is to dump the coverage report in the file 
 # --cov PATH is the directory PATH to consider for coverage. Default to SRC_DIR=src
 # --cov-fail-under=100 is to make the command fail if coverage does not reach a 100%
-# --cov-report=term-missing:skip-covered is used to print the missing lines for coverage withtout 
-# taking into account skiped tests
+# --cov-report=term-missing:skip-covered is used to avoid printing covered lines for all files
 .PHONY: pytest # Run pytest on all tests
 pytest:
 	"$(MAKE)" pytest_internal_parallel \
@@ -273,6 +272,8 @@ pytest_no_flaky: check_current_flaky_tests
 # --cov PATH is the directory PATH to consider for coverage. Default to SRC_DIR=src
 # --cov-append is to make the coverage of the previous pytest run to also consider the tests that are
 # going to be re-executed by 'pytest_run_last_failed'
+# --cov-fail-under=100 is to make the command fail if coverage does not reach a 100%
+# --cov-report=term-missing:skip-covered is used to avoid printing covered lines for all files
 # --global-coverage-infos-json=global-coverage-infos.json is to dump the coverage report in the file 
 # --last-failed runs all last failed tests
 # --last-failed-no-failures none' indicates pytest not to run anything (instead of running 
@@ -282,6 +283,8 @@ pytest_run_last_failed:
 	poetry run pytest $(TEST) \
 	--cov=$(SRC_DIR) \
 	--cov-append \
+	--cov-fail-under=100 \
+	--cov-report=term-missing:skip-covered \
 	--global-coverage-infos-json=global-coverage-infos.json \
 	--last-failed \
 	--last-failed-no-failures none

script/actions_utils/pytest_failed_test_report.py

@@ -97,8 +97,14 @@ def write_failed_tests_report(
                 else:
                     failed_tests_report["non_flaky"].append(test_name)  # type: ignore[attr-defined]
 
-        # If no non-flaky tests failed, report that all failed tests were known flaky tests
-        if not failed_tests_report["non_flaky"]:
+        # If there are some flaky tests but no non-flaky tests failed, report that all failed tests
+        # were known flaky tests
+        # We need to make sure that at least one flaky test has been detected for one specific
+        # reason: if, for example, a test file has a syntax error, pytest will "crash" and therefore
+        # won't collect any tests in the file. The problem is that this will return an 'exitcode'
+        # of 1, making this script unexpectedly return 'all_failed_tests_are_flaky=True' in the
+        # case where 'failed_tests_report["non_flaky"]' is empty
+        if failed_tests_report["flaky"] and not failed_tests_report["non_flaky"]:
             failed_tests_report["all_failed_tests_are_flaky"] = True
 
     else: