HSEARCH-3739 Use a ThreadPool instead of a ForkJoinPool for asynchron…

…ous mass indexing

The ForkJoinPool does not use interruption to cancel tasks, which makes
it hard to integrate to the current implementation of MassIndexer where
everything uses interruptions for cancellation.

mapper/orm/src/main/java/org/hibernate/search/mapper/orm/massindexing/impl/MassIndexerImpl.java

@@ -11,7 +11,7 @@
 import java.util.LinkedHashSet;
 import java.util.Set;
 import java.util.concurrent.CompletableFuture;
-import java.util.concurrent.ForkJoinPool;
+import java.util.concurrent.ExecutorService;
 
 import org.hibernate.CacheMode;
 import org.hibernate.search.engine.backend.session.spi.DetachedBackendSessionContext;
@@ -164,7 +164,16 @@ public MassIndexer monitor(MassIndexingMonitor monitor) {
 
 	@Override
 	public CompletableFuture<?> start() {
-		return Futures.runAsync( createCoordinator(), ForkJoinPool.commonPool() );
+		BatchCoordinator coordinator = createCoordinator();
+		ExecutorService executor = mappingContext.getThreadPoolProvider()
+				.newFixedThreadPool( 1, THREAD_NAME_PREFIX + "Coordinator" );
+		try {
+			return Futures.runAsync( coordinator, executor );
+		}
+		finally {
+			executor.shutdown();
+		}
+
 	}
 
 	@Override