Exclude deleted documents when extracting stored fields

When reading stored fields, `loadDocument` calls
`IndexReader.storedFields().document(id, ...)` to get back the stored
fields for a given document.  However, Lucene doesn't automatically
check that the requested document is still live:

> NOTE: for performance reasons, this method does not check if the
> requested document is deleted, and therefore asking for a deleted
> document may yield unspecified results. Usually this is not
> required, however you can test if the doc is deleted by checking the
> Bits returned from
> MultiBits.getLiveDocs(org.apache.lucene.index.IndexReader).

This can lead to browse headings appearing that don't actually link to
anything.

Adds a check against the live docs bitset, as described in the
documentation above.

browse-indexing/StoredFieldLeech.java

@@ -6,6 +6,7 @@
 import org.apache.lucene.store.*;
 import org.apache.lucene.index.*;
 import org.apache.lucene.document.*;
+import org.apache.lucene.util.Bits;
 
 import org.vufind.util.Utils;
 import org.vufind.util.BrowseEntry;
@@ -20,6 +21,7 @@ public class StoredFieldLeech extends Leech
 
     private Set<String> fieldSelection;
 
+    private Bits liveDocsBitSet;
 
     public StoredFieldLeech(String indexPath, String field) throws Exception
     {
@@ -40,6 +42,10 @@ public StoredFieldLeech(String indexPath, String field) throws Exception
         fieldSelection.add("id");   // make Solr id available for error messages
 
         reader = DirectoryReader.open(FSDirectory.open(new File(indexPath).toPath()));
+
+        // Will be null if the index contains no deletes.
+        liveDocsBitSet = MultiBits.getLiveDocs(reader);
+
         buffer = new LinkedList<BrowseEntry> ();
     }
 
@@ -80,7 +86,9 @@ public BrowseEntry next() throws Exception
     {
         while (buffer.isEmpty()) {
             if (currentDoc < reader.maxDoc()) {
-                loadDocument(reader, currentDoc);
+                if (this.liveDocsBitSet == null || this.liveDocsBitSet.get(currentDoc)) {
+                    loadDocument(reader, currentDoc);
+                }
                 currentDoc++;
             } else {
                 return null;