[ML] Increase ScrollDataExtractor context duration

In cases where the job is bound on the analytics performance
the datafeed can fail because the scroll expires. This is
commit increases the scroll context duration from 10 to 30 minutes
as a temporary solution that will avoid most cases.

Original commit: elastic/x-pack-elasticsearch@fd277bbaa1

plugin/src/main/java/org/elasticsearch/xpack/ml/datafeed/extractor/scroll/ScrollDataExtractor.java

@@ -43,7 +43,7 @@ import java.util.concurrent.TimeUnit;
 class ScrollDataExtractor implements DataExtractor {
 
     private static final Logger LOGGER = Loggers.getLogger(ScrollDataExtractor.class);
-    private static final TimeValue SCROLL_TIMEOUT = new TimeValue(10, TimeUnit.MINUTES);
+    private static final TimeValue SCROLL_TIMEOUT = new TimeValue(30, TimeUnit.MINUTES);
 
     private final Client client;
     private final ScrollDataExtractorContext context;