Fixes broken metrics unit tests

Quite often unit test failure showing "*** a different view
with the same name is already registered" messages at many places.
At the biggning of every metrics test, it tries to register respective
metrics views(metrics data it want present), publish metrics and
uniregister views. Uniregister view is important becuase for next
test it dont need to hold the old test data.
unregister() views were getting called using `defer` in
table test for loop, just like
for _, td := range tests {
	defer unregister()
	.......
}
This defer actually gets execute after completion of test function rather
than for each loop. This were resulting unwanted execution behaviour
for unregister() views function, which umtimately faling tests.

This patch fixes the behaviour by calling uniregister() views function
at the end of loop.

Fixes tektoncd#1659

pkg/reconciler/pipelinerun/metrics_test.go

@@ -93,7 +93,6 @@ func TestRecordPipelineRunDurationCount(t *testing.T) {
 
 	for _, test := range testData {
 		t.Run(test.name, func(t *testing.T) {
-			defer unregisterMetrics()
 
 			metrics, err := NewRecorder()
 			assertErrIsNil(err, "Recorder initialization failed", t)
@@ -102,6 +101,8 @@ func TestRecordPipelineRunDurationCount(t *testing.T) {
 			assertErrIsNil(err, "DurationAndCount recording recording got an error", t)
 			metricstest.CheckDistributionData(t, "pipelinerun_duration_seconds", test.expectedTags, 1, test.expectedDuration, test.expectedDuration)
 			metricstest.CheckCountData(t, "pipelinerun_count", test.expectedTags, test.expectedCount)
+
+			unregisterMetrics()
 		})
 	}
 }
@@ -121,6 +122,7 @@ func TestRecordRunningPipelineRunsCount(t *testing.T) {
 	err = metrics.RunningPipelineRuns(informer.Lister())
 	assertErrIsNil(err, "RunningPrsCount recording expected to return nil but got error", t)
 	metricstest.CheckLastValueData(t, "running_pipelineruns_count", map[string]string{}, 1)
+
 }
 
 func addPipelineRun(informer alpha1.PipelineRunInformer, run, pipeline, ns string, status corev1.ConditionStatus, t *testing.T) {

pkg/reconciler/taskrun/metrics_test.go

@@ -101,15 +101,15 @@ func TestRecordTaskrunDurationCount(t *testing.T) {
 
 	for _, test := range testData {
 		t.Run(test.name, func(t *testing.T) {
-			defer unregisterMetrics()
-
 			metrics, err := NewRecorder()
 			assertErrIsNil(err, "Recorder initialization failed", t)
 
 			err = metrics.DurationAndCount(test.taskRun)
 			assertErrIsNil(err, "DurationAndCount recording got an error", t)
 			metricstest.CheckDistributionData(t, "taskrun_duration_seconds", test.expectedTags, 1, test.expectedDuration, test.expectedDuration)
 			metricstest.CheckCountData(t, "taskrun_count", test.expectedTags, test.expectedCount)
+
+			unregisterMetrics()
 		})
 	}
 }
@@ -179,15 +179,15 @@ func TestRecordPipelinerunTaskrunDurationCount(t *testing.T) {
 
 	for _, test := range testData {
 		t.Run(test.name, func(t *testing.T) {
-			defer unregisterMetrics()
-
 			metrics, err := NewRecorder()
 			assertErrIsNil(err, "Recorder initialization failed", t)
 
 			err = metrics.DurationAndCount(test.taskRun)
 			assertErrIsNil(err, "DurationAndCount recording got an error", t)
 			metricstest.CheckDistributionData(t, "pipelinerun_taskrun_duration_seconds", test.expectedTags, 1, test.expectedDuration, test.expectedDuration)
 			metricstest.CheckCountData(t, "taskrun_count", test.expectedTags, test.expectedCount)
+
+			unregisterMetrics()
 		})
 	}
 }
@@ -255,8 +255,6 @@ func TestRecordPodLatency(t *testing.T) {
 
 	for _, td := range testData {
 		t.Run(td.name, func(t *testing.T) {
-			defer unregisterMetrics()
-
 			metrics, err := NewRecorder()
 			assertErrIsNil(err, "Recorder initialization failed", t)
 
@@ -267,6 +265,8 @@ func TestRecordPodLatency(t *testing.T) {
 			}
 			assertErrIsNil(err, "RecordPodLatency recording expected to return nil but got error", t)
 			metricstest.CheckLastValueData(t, "taskruns_pod_latency", td.expectedTags, td.expectedValue)
+
+			unregisterMetrics()
 		})
 	}