use loose tag parsing in output corpuses in data rule tests

tests/datarule.cpp

@@ -102,6 +102,7 @@ void test_one_rule_item_actual(const rule_compare_test& c)
 		BOOST_ERROR("Empty chunk loaded");
 	}
 	Corpus2::XcesReader xr(tagset, c.output_corpus.string());
+	xr.set_option("loose");
 	boost::shared_ptr<Corpus2::Chunk> expected = xr.get_next_chunk();
 	BOOST_REQUIRE_EQUAL(chunk->sentences().size(), expected->sentences().size());
 	Wccl::Parser parser(tagset);