[Backport 7.9] Make query results more resilient to not break on unauthorized results but skip them

When iterating through the nodes in a query result via

javax.jcr.query.QueryResult queryResult = ...;
for (Node node : new NodeIterable(queryResult.getNodes())) {
        System.out.println(node.getPath());
}

It will always work since in case a node in the search result is unauthorized (should normally not happen), it still gets skipped by org.apache.jackrabbit.core.query.lucene.NodeIteratorImpl (and only a warning is logged). Since search results are non-transactional, this (an unauthorized node during iteration being skipped) can always happen even without the authorization query.

If however, instead of queryResult.getNodes(), you use

javax.jcr.query.QueryResult queryResult = ...;
 while (rows.hasNext()) {
         final Row row = rows.nextRow();
         final Value[] values = row.getValues();
}

the code above throws a repository exception on 'row.getValues()' if the backing jcr node cannot be read (unauthorized).
We should in jackrabbit RowIteratorImpl first fetch the node and log a warning if it is not allowed to be read.