Make encoding detection engine public (KEncodingDetector) part of kdecore.
Now, together with structurized KCodecAction, it can be used in any app to provide encoding detection magic to the average user.
(this, however, mainly defines interface, and the actual heuristics is subject for the improvement)

Integrate this feature into katepart and back into khtmlpart.
Now go to katepart open/save configuration page and make you choice!

Improve detection heuristics for Cyrillic and Western European (almost 100%!) along the way

TODO: It already checks for html and xml tags, so check also for \usepackage[utf8]{inputenc}

Maybe it will take advantage of 150kb libenca someday, when its cyr detection heuristics become more sophisticated and the humans get terabytes of RAM in their brains^Whomes

Fixing anlog the way palindromic bug 55355