addressed PR feedback

Author: bachuv

client/src/main/java/com/microsoft/durabletask/DurableTaskClient.java

@@ -311,7 +311,8 @@ public void rewindInstance(String instanceId) {
      * <p>
      * This method can only be used on orchestration instances that are in a <code>Failed</code> state.
      * When rewound, the orchestration instance will restart from the point of failure as if the failure
-     * never occurred.
+     * never occurred. It rewinds the orchestration by replaying any
+     * Failed Activities and Failed suborchestrations that themselves have Failed Activities
      *
      * @param instanceId the ID of the orchestration instance to rewind
      * @param reason the reason for rewinding the orchestration instance

client/src/test/java/com/microsoft/durabletask/IntegrationTests.java

@@ -419,6 +419,8 @@ void subOrchestrationWithFailedActivity() throws TimeoutException {
         final String childOrchestratorName = "ChildOrchestration";
         final String activityName = "FailingActivity";
         final String failureMessage = "Simulated activity failure in sub-orchestration";
+        final String expectedOutput = "Success after rewind";
+        final AtomicBoolean shouldFail = new AtomicBoolean(true);
 
         DurableTaskGrpcWorker worker = this.createWorkerBuilder()
                 .addOrchestrator(parentOrchestratorName, ctx -> {
@@ -430,20 +432,34 @@ void subOrchestrationWithFailedActivity() throws TimeoutException {
                     ctx.complete(result);
                 })
                 .addActivity(activityName, ctx -> {
-                    throw new RuntimeException(failureMessage);
+                    if (shouldFail.compareAndSet(true, false)) {
+                        throw new RuntimeException(failureMessage);
+                    }
+                    return expectedOutput;
                 })
                 .buildAndStart();
 
         DurableTaskClient client = this.createClientBuilder().build();
         try (worker; client) {
             String instanceId = client.scheduleNewOrchestrationInstance(parentOrchestratorName);
+
+            // Wait for the orchestration to fail due to the activity failure in the sub-orchestration
             OrchestrationMetadata instance = client.waitForInstanceCompletion(instanceId, defaultTimeout, true);
             assertNotNull(instance);
             assertEquals(OrchestrationRuntimeStatus.FAILED, instance.getRuntimeStatus());
 
             FailureDetails details = instance.getFailureDetails();
             assertNotNull(details);
             assertTrue(details.getErrorMessage().contains(failureMessage));
+
+            // Rewind the failed orchestration
+            client.rewindInstance(instanceId, "Rewinding sub-orchestration with failed activity");
+
+            // Wait for the orchestration to complete after rewind
+            instance = client.waitForInstanceCompletion(instanceId, defaultTimeout, true);
+            assertNotNull(instance);
+            assertEquals(OrchestrationRuntimeStatus.COMPLETED, instance.getRuntimeStatus());
+            assertEquals(expectedOutput, instance.readOutputAs(String.class));
         }
     }