About KR: I don't quite understand, why it is allowed for a player to reset his or hers KR. If you even sometimes use KR to check balances, it should not be allowed. Certain people reset several times when they are playing. Also never reseting is a problem since then KR won't move much, and skilled players can end up with much lower KR than their real level. I recommend either timed resets (like once a month) or resets by request done by an admin.
Regardless, when used properly KR does mesure what it is supposed to measure. That is kill to death ratio, and that only. Killing bots only won't raise it too much. If there is an issue with it, it is people who play on public servers as if they are some kind of ETPRO games where you don't even try to progress the obj. Long killing sprees are what raise KR the most and those you don't really need to complete objectives.
My solutions:
1) Remove KR reseting ability. Howfair command should be optimally used around 1m after map starts unless admin is able to otherwise check teams just by looking at player names. Servers auto shuffle at the start so sometimes players stay in spec until map has properly started.
2) Use few different shuffling (both auto and manual) techniques. When shuffle doesn't work it would be useful to have a different shuffling mechanism. Perhaps a combination of battle sense, XP gathered during last 1h or so, class related XP (medic, engi etc.) or something else. This way you can get different teams after each shuffle.
3) Have some kind of general rules when /scores (kill difference) are unfair. Is 200 kill difference enough?
4) Pay more attention to map progress. I believe teams in public games should be made such that attacking team should always have a fair shot at winning the game. Games are much more interesting when attackers are slightly stronger (not that it should be the goal) and other way around.
I realize some of these require a bit of coding. I don't expect them to be implemented just like that. But please, if you reply, please but some effort in thinking why my ideas might be good or bad.