Replace caplog with patching log property in k8s tests

Since caplog is very vulnerable to side effects we remove it from
k8s tests and replace it with patching the log property of the
K8SHook. This should not be vulnerable to side-effects.

providers/tests/cncf/kubernetes/hooks/test_kubernetes.py

@@ -976,28 +976,33 @@ async def test_delete_pod(self, lib_method, kube_config_loader):
 
     @pytest.mark.asyncio
     @mock.patch(KUBE_API.format("read_namespaced_pod_log"))
-    async def test_read_logs(self, lib_method, kube_config_loader, caplog):
+    async def test_read_logs(self, lib_method, kube_config_loader):
         lib_method.return_value = self.mock_await_result("2023-01-11 Some string logs...")
-
         hook = AsyncKubernetesHook(
             conn_id=None,
             in_cluster=False,
             config_file=None,
             cluster_context=None,
         )
-        await hook.read_logs(
-            name=POD_NAME,
-            namespace=NAMESPACE,
-        )
+        with mock.patch(
+            "airflow.providers.cncf.kubernetes.hooks.kubernetes.AsyncKubernetesHook.log",
+            new_callable=PropertyMock,
+        ) as log:
+            await hook.read_logs(
+                name=POD_NAME,
+                namespace=NAMESPACE,
+            )
 
-        lib_method.assert_called_once()
-        lib_method.assert_called_with(
-            name=POD_NAME,
-            namespace=NAMESPACE,
-            follow=False,
-            timestamps=True,
-        )
-        assert "Container logs from 2023-01-11 Some string logs..." in caplog.text
+            lib_method.assert_called_once()
+            lib_method.assert_called_with(
+                name=POD_NAME,
+                namespace=NAMESPACE,
+                follow=False,
+                timestamps=True,
+            )
+            log.return_value.info.assert_called_with(
+                "Container logs from %s", "2023-01-11 Some string logs..."
+            )
 
     @pytest.mark.asyncio
     @mock.patch(KUBE_BATCH_API.format("read_namespaced_job_status"))