fix performance regression

The last improvement of memory usage introduced a performance
regression. The ingestion performance dropped by 50%-80%, because
for every inserted entry the Tags were created inefficient.

pdb-api/src/test/java/org/lucares/memory/MemoryScale.java

@@ -9,6 +9,7 @@ import java.util.Map;
 import org.lucares.pdb.api.StringCompressor;
 import org.lucares.pdb.api.Tag;
 import org.lucares.pdb.api.Tags;
+import org.lucares.pdb.api.TagsBuilder;
 import org.lucares.pdb.api.UniqueStringIntegerPairs;
 
 public class MemoryScale {
@@ -70,7 +71,6 @@ public class MemoryScale {
 		default:
 			return null;
 		}
-
 	}
 
 	private static Object createTag() {
@@ -90,13 +90,14 @@ public class MemoryScale {
 	}
 
 	private static Object createTags6() {
-		Tags result = Tags.create("k1", "v1");
-		result = result.copyAdd("k2", "v2");
-		result = result.copyAdd("k3", "v3");
-		result = result.copyAdd("k4", "v4");
-		result = result.copyAdd("k5", "v5");
-		result = result.copyAdd("k6", "v6");
-		return result;
+		TagsBuilder result = TagsBuilder.create();
+		result = result.add("k1", "v1");
+		result = result.add("k2", "v2");
+		result = result.add("k3", "v3");
+		result = result.add("k4", "v4");
+		result = result.add("k5", "v5");
+		result = result.add("k6", "v6");
+		return result.build();
 	}
 
 	private static Object createPathAsUtf8(final String string) {