Improve performance of PathRetractionStrategy

Helpful for traversals with lots of children where labels don't need to propagate. CTR

CHANGELOG.asciidoc

@@ -25,6 +25,7 @@ image::https://raw.githubusercontent.com/apache/tinkerpop/master/docs/static/ima
 
 * Fixed a bug in Gremlin.Net for .NET 8 that led to exceptions: `InvalidOperationException: Enumeration has not started. Call MoveNext.`
 * Fixed message requestId serialization in `gremlin-python`.
+* Improved performance of `PathRetractionStrategy` for traversals that carry many children, but don't hold many labels to propogate.
 * Fixed bug in bytecode translation of `g.tx().commit()` and `g.tx().rollback()` in all languages.
 * Improved error message from `JavaTranslator` by including exception source.
 * Added missing `short` serialization (`gx:Int16`) to GraphSONV2 and GraphSONV3 in `gremlin-python`.

gremlin-core/src/main/java/org/apache/tinkerpop/gremlin/process/traversal/strategy/optimization/PathRetractionStrategy.java

@@ -249,6 +249,10 @@ private static boolean isNotApplicable(final Traversal.Admin<?, ?> traversal) {
     }
 
     private void applyToChildren(final Set<String> keepLabels, final List<Traversal.Admin<Object, Object>> children) {
+        // if there are no labels to keep, then there no need to iterate all the children because we won't be
+        // adding anything PathProcessor keepLabels - avoids the added recursion
+        if (keepLabels.isEmpty()) return;
+
         for (final Traversal.Admin<Object, Object> child : children) {
             TraversalHelper.applyTraversalRecursively(trav -> addLabels(trav, keepLabels), child);
         }