Instead of open a document for every entry we do open it only once per
search. For that we sort the entries by file path, so that they are clustered
together and reuse the last document if the path hasn't changed. This can
improve the calculations of the search results drastically.
Change-Id: I9c9c1e387624297d84c6a2ca6edb6130f739d295
Reviewed-by: Nikolai Kosjar <nikolai.kosjar@qt.io>
