This fixes a few smaller issues in the code that parses and interprets the queries found when looking into long-running requests on our production servers. It mainly makes searches a tiny bit faster.
avoid duplicates in the token lists used in SQL queries
try to avoid index use on the address index against very frequent tokens
category words should not appear in the middle or a query string or be used together with housenumber-only searches
make sure penalties for housenumber-only searches are comparable with the equivalent named address search
This fixes a few smaller issues in the code that parses and interprets the queries found when looking into long-running requests on our production servers. It mainly makes searches a tiny bit faster.