Use pytest-repeat to reproduce failure on CI.

.github/actions/run-python-test-set/action.yml

@@ -19,7 +19,7 @@ inputs:
   run_in_parallel:
     description: 'Whether to run tests in parallel'
     required: false
-    default: 'true'
+    default: 'false'
   save_perf_report:
     description: 'Whether to upload the performance report, if true PERF_TEST_RESULT_CONNSTR env variable should be set'
     required: false
@@ -171,7 +171,7 @@ runs:
           --junitxml=$TEST_OUTPUT/junit.xml \
           --alluredir=$TEST_OUTPUT/allure/results \
           --tb=short \
-          --verbose \
+          --verbose -k "test_forward or test_create_snapsh" -x \
           -rA $TEST_SELECTION $EXTRA_PARAMS
 
         if [[ "${{ inputs.save_perf_report }}" == "true" ]]; then

.github/workflows/build_and_test.yml

@@ -5,6 +5,7 @@ on:
     branches:
       - main
       - release
+      - tmp-repro
   pull_request:
 
 defaults: