Added code comments

Signed-off-by: Martin Gaievski <gaievski@amazon.com>

CHANGELOG.md

@@ -17,7 +17,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 ### Features
 ### Enhancements
 ### Bug Fixes
-- Fixed merge logic in hybrid query for multiple shards case ([#876](https://github.com/opensearch-project/neural-search/pull/876))
+- Fixed merge logic in hybrid query for multiple shards case ([#877](https://github.com/opensearch-project/neural-search/pull/877))
 ### Infrastructure
 - Update batch related tests to use batch_size in processor & refactor BWC version check ([#852](https://github.com/opensearch-project/neural-search/pull/852))
 ### Documentation

src/main/java/org/opensearch/neuralsearch/search/query/TopDocsMerger.java

@@ -55,6 +55,9 @@ class TopDocsMerger {
      * @return merged TopDocsAndMaxScore object
      */
     public TopDocsAndMaxScore merge(final TopDocsAndMaxScore source, final TopDocsAndMaxScore newTopDocs) {
+        // we need to check if any of source and destination top docs are empty. This is needed for case when concurrent segment search
+        // is enabled. In such case search is done by multiple workers, and results are saved in multiple doc collectors. Any on those
+        // results can be empty, in such case we can skip actual merge logic and just return result object.
         if (isEmpty(newTopDocs)) {
             return source;
         }