HHH-19089 lower collection pre-sizing

see hibernate#9683 (comment)

this can avoid memory peaks and CPU spikes because of heavy GC when setMaxResults is called with a very large limit e.g. Integer.MAX_VALUE (probably accidentally). it turned out the the previous limit of 1million was too large and could cause issues in production systems where legacy code bases where not aware of such collection pre-sizing

Author: chrisrueger

hibernate-core/src/main/java/org/hibernate/sql/results/spi/ListResultsConsumer.java

@@ -33,10 +33,10 @@
 public class ListResultsConsumer<R> implements ResultsConsumer<List<R>, R> {
 
 	/**
-	 * Let's be reasonable: a row estimate greater than 1M rows is probably either a misestimate or a bug,
-	 * so let's set {@code 2^20} which is a bit above 1M as maximum collection size.
+	 * Let's be reasonable: a row estimate greater than 8k rows is probably either a misestimate or a bug,
+	 * so let's set {@code 2^13} which is a bit above 8k as maximum collection size.
 	 */
-	private static final int INITIAL_COLLECTION_SIZE_LIMIT = 1 << 20;
+	private static final int INITIAL_COLLECTION_SIZE_LIMIT = 1 << 13;
 
 	private static final ListResultsConsumer<?> NEVER_DE_DUP_CONSUMER = new ListResultsConsumer<>( UniqueSemantic.NEVER );
 	private static final ListResultsConsumer<?> ALLOW_DE_DUP_CONSUMER = new ListResultsConsumer<>( UniqueSemantic.ALLOW );