When running a query that involves equality matching (fixed size index) and ordered matching (unknown size) prefer the shortest index to speed up query time.
Bug - based on the alphabetical ordering, the shortest index might not be chosen, causing a performance hit.
e.g. "content = '5' AND category <= 1000" content should be chosen, but because alphabetically 'co' is before 'ca' then category is searched first, and
since it is of unlimited size, the shortest scan is skipped.
A Customer need to backport to 14.01 and 14.5