Leaving it -- and all skill things like it -- out entirely, dealing with it in a reasonable way between player and DM, makes the game more fun and improves the quality and value of play.
Or you can have siloing. RM, in effect, has this, because PCs have many development points per level, and are constrained in how many they can spend each level on "core" skills like combat and magic.
Or, like (I think) Ariosto has been saying, you can do it like RQ does and silo every single skill,
and determine progression in each skill independently by a "skill gain" roll after each use of a skill.
And interestingly, these different ways of doing it have different effects on the game. RM favours heavy metagaming of PC build, whether for combat/magic optimisation, or flavour "optimisation" (goat herding, fire building and similar peripheral skills purchased out of the siloed development points). Whereas RQ is almost metagame free, because there is nothing that affects skill development except ingame use.
It seems to me that there is another dispute between Hussar and Ariosto, also, namely, whether a richer skill system or a richer action resolution system (and these two can overlap but need not) reduces the need to "game the GM". I feel that even a system like 3E or RM can still lead to a degree of GM-gaming, because there aren't tables or preset DCs to cover every eventuality, and sometimes the GM just has to make a call. I prefer more expressly metagame-y systems, like DMG p 42, for reining this in. (I'm pretty sure from past exchanges that Ariosto doesn't agree with this.) They still depend upon reasonable give-and-take between players and GM, but it's within a pre-established framework. So players can have a greater degree of confidence in the consistency and tenability of the way that declared actions are resolved.
I think you're throwing the baby out with the bathwater. Why not have an "Outdoor Adventurers" stat? This mechanic gives you your relative skill at doing anything outdoorsey that is within reason. Depending on the level of the stat vs the difficulty of the job, you get a mechanic to distinguish my character from yours. Maybe my fighter is a knight who always had a squire. What do I know about cooking rabbits? I had servants for that.
For me, the issue is "In what way does cooking rabbits matter to the play of the game?" If not at all, then why clutter the character sheet and the rulebooks? On the other hand, if it's a wilderness survival game, maybe those sorts of skills should be front-and-centre. If (as is often suggested in relation to D&D, and especially Craft and Profession in 3E) it's about establishing character background, then in many cases it should be possible just to make a note and move on. Or, in a system like RM in which the skill list purports to be the totality of the character, you buy it out of the relevant siloed character building resources.
Importantly, even if it's just a note written on a character sheet it can still matter to play. For example, in resolving a wilderness survival skill challenge in 4e, I would expect the knight without his squire to approach the scenario, and to deploy his skills, in quite a different way from a ranger trained in Nature. So it would still matter to the unfolding story. (My feeling is that in a more strongly simulationist action resolution system its harder to make mere notes written on character sheets actually matter in this way.)