Make sure that all unittests are executed (triton-inference-server#1823)

* Verify execution of unittests

* Add comments to InferUnit

* Improve test.sh to work better in case of failures

* improve the error messages

* rename the unittests base class and fix minor issues

* [WIP] Add counting successful runs to other testcases

* Update all "test.sh" files to make it compatible with new test counting

* fix a minor bug in test counting

* fix whitespace

* fix minor bug in check_test_results

* fix a minor bug in check_test_results

* fix minor bugs in test counting

* fix a minor bug in L0_infer

* fix minor bug in L0_infer_shm

* rename TestResCollector -> TestResultCollector

* L0_infer TestResult

* Make test counting more robust

Author: Tabrizian

.gitignore

@@ -1,2 +1,3 @@
 /builddir
 /.vscode
+*.so

qa/L0_batcher/batcher_test.py

@@ -59,7 +59,7 @@
 _deferred_exceptions = []
 
 
-class BatcherTest(unittest.TestCase):
+class BatcherTest(tu.TestResultCollector):
 
     def setUp(self):
         # The helper client for setup will be GRPC for simplicity.
@@ -77,6 +77,7 @@ def tearDown(self):
             destroy_fun = cudashm.destroy_shared_memory_region
         for precreated_shm_region in self.precreated_shm_regions_:
             destroy_fun(precreated_shm_region)
+        super().tearDown()
 
     # FIXME why only used for outputs
     def create_advance(self, shm_regions=None):