Fix bugs when using a custom FailureConverter in tests

Follow up to temporalio#1484.

1. We weren't correctly threading the failure converter down to
   activities and child workflows, so they were still using the
   default converter.

2. The `NonRetryable` flag on the protobuf failure type wasn't being
   honored for custom errors.

internal/internal_workflow_testsuite.go

@@ -374,6 +374,7 @@ func (env *testWorkflowEnvironmentImpl) newTestWorkflowEnvironmentForChild(param
 	childEnv.testWorkflowEnvironmentShared = env.testWorkflowEnvironmentShared
 	childEnv.workerOptions = env.workerOptions
 	childEnv.dataConverter = params.DataConverter
+	childEnv.failureConverter = env.failureConverter
 	childEnv.registry = env.registry
 	childEnv.detachedChildWaitDisabled = env.detachedChildWaitDisabled
 
@@ -1403,8 +1404,14 @@ func (env *testWorkflowEnvironmentImpl) executeActivityWithRetryForTest(
 
 		// check if a retry is needed
 		if request, ok := result.(*workflowservice.RespondActivityTaskFailedRequest); ok && parameters.RetryPolicy != nil {
+			failure := request.GetFailure()
+
+			if failure.GetApplicationFailureInfo().GetNonRetryable() {
+				break
+			}
+
 			p := fromProtoRetryPolicy(parameters.RetryPolicy)
-			backoff := getRetryBackoffWithNowTime(p, task.GetAttempt(), env.failureConverter.FailureToError(request.GetFailure()), env.Now(), expireTime)
+			backoff := getRetryBackoffWithNowTime(p, task.GetAttempt(), env.failureConverter.FailureToError(failure), env.Now(), expireTime)
 			if backoff > 0 {
 				// need a retry
 				waitCh := make(chan struct{})
@@ -1987,6 +1994,7 @@ func (env *testWorkflowEnvironmentImpl) newTestActivityTaskHandler(taskQueue str
 		MetricsHandler:     env.metricsHandler,
 		Logger:             env.logger,
 		UserContext:        env.workerOptions.BackgroundActivityContext,
+		FailureConverter:   env.failureConverter,
 		DataConverter:      dataConverter,
 		WorkerStopChannel:  env.workerStopChannel,
 		ContextPropagators: env.contextPropagators,