[fast-client] Add metric for batchGet requests routed to single-GET lookups

Add a new fast-client metric, batch_get_routed_to_single_get_request_count
(OccurrenceRate on the MULTI_GET_STREAMING stats), recorded at the single-key
short-circuit choke point
(InternalAvroStoreClient#streamingBatchGetForSingleKey)
so it counts every 1-key batchGet/streamingBatchGet served via a single-GET
lookup,
regardless of which layer (Retriable or Dispatching) triggered the
short-circuit.

Extend testSingleKeyStreamingBatchGetUsesSingleGet to assert the metric is
recorded.

Author: Copilot

clients/venice-client/src/main/java/com/linkedin/venice/fastclient/InternalAvroStoreClient.java

@@ -8,6 +8,7 @@
 import com.linkedin.venice.client.store.streaming.VeniceResponseMap;
 import com.linkedin.venice.client.store.streaming.VeniceResponseMapImpl;
 import com.linkedin.venice.compute.ComputeRequestWrapper;
+import com.linkedin.venice.read.RequestType;
 import com.linkedin.venice.utils.concurrent.VeniceConcurrentHashMap;
 import java.util.Map;
 import java.util.Optional;
@@ -52,6 +53,7 @@ public CompletableFuture<V> get(K key) throws VeniceClientException {
    * long-tail retry on 1-key batch gets.
    */
   protected final void streamingBatchGetForSingleKey(K key, StreamingCallback<K, V> callback) {
+    getClientConfig().getStats(RequestType.MULTI_GET_STREAMING).recordBatchGetRoutedToSingleGetRequest();
     get(key).whenComplete((value, throwable) -> {
       if (throwable != null) {
         callback.onCompletion(Optional.of(toException(throwable)));

clients/venice-client/src/main/java/com/linkedin/venice/fastclient/stats/FastClientStats.java

@@ -50,6 +50,8 @@ public class FastClientStats extends ClientStats {
   private final Sensor dualReadThinClientFastClientLatencyDeltaSensor;
 
   private final Sensor leakedRequestCountSensor;
+  /** Counts batch-get requests that contained a single key and were therefore served via a single-GET lookup. */
+  private final Sensor batchGetRoutedToSingleGetRequestCountSensor;
   private volatile MetricEntityStateOneEnum<RejectionReason> rejectionRatio;
 
   // OTel metrics
@@ -98,6 +100,8 @@ private FastClientStats(MetricsRepository metricsRepository, String storeName, R
     this.dualReadThinClientFastClientLatencyDeltaSensor =
         registerSensorWithDetailedPercentiles("dual_read_thinclient_fastclient_latency_delta", new Max(), new Avg());
     this.leakedRequestCountSensor = registerSensor("leaked_request_count", new OccurrenceRate());
+    this.batchGetRoutedToSingleGetRequestCountSensor =
+        registerSensor("batch_get_routed_to_single_get_request_count", new OccurrenceRate());
   }
 
   /**
@@ -270,6 +274,14 @@ public void recordRejectionRatio(double rejectionRatio) {
     this.rejectionRatio.record(rejectionRatio, RejectionReason.THROTTLED_BY_LOAD_CONTROLLER);
   }
 
+  /**
+   * Records a batch-get/streaming-batch-get request that contained a single key and was therefore short-circuited
+   * to a single-GET lookup (see {@code InternalAvroStoreClient#streamingBatchGetForSingleKey}).
+   */
+  public void recordBatchGetRoutedToSingleGetRequest() {
+    batchGetRoutedToSingleGetRequestCountSensor.record();
+  }
+
   /**
    * This method is a utility method to build concise summaries useful in tests
    * and for logging. It generates a single string for all metrics for a sensor

clients/venice-client/src/test/java/com/linkedin/venice/fastclient/BatchGetAvroStoreClientUnitTest.java

@@ -89,6 +89,15 @@ public void testSingleKeyStreamingBatchGetUsesSingleGet()
         client.getSimulatorCompleteFuture());
 
     validateMetrics(client, 1, 1, 0, 0);
+
+    // The single-key batch-get should be counted as routed to a single-GET lookup.
+    Map<String, ? extends Metric> metrics = getStats(client.getClientConfig());
+    assertTrue(
+        metrics
+            .get(
+                "." + client.UNIT_TEST_STORE_NAME
+                    + "--multiget_streaming_batch_get_routed_to_single_get_request_count.OccurrenceRate")
+            .value() > 0);
   }
 
   /**