Concerning the relative usefulness of snap percentage versus targets:
How much "autonomy" (for lack of a better word) do QBs have when throwing the ball? My understanding is that a play is more than just a set of routes, its also a set progression the QB goes through looking for an open guy. So if you're a TE, say, and you play on 80% of snaps and run routes on most snaps, but you're always the 4th or 5th option, it doesn't matter how many snaps you see and it's not the QBs fault for not throwing you the ball ... you simply aren't part of the game plan.
On the other hand there are QBs that will stare a guy down and force the ball in even if he's not open (I'm looking at you, Vick). This is going to result in a lot of targets but not many catches, so maybe targets aren't a great stat to use for those guys.
So it seems that what's important is highly dependent on what team and QB you're looking at. Let's say a team has a rookie QB and they're trying to take it easy on him by limiting the number of reads he makes during a play. If a WR sees a lot of snaps but he's the 3rd or 4th option, the fact that he has a high snap percentage is probably meaningless. For a veteran QB who's good at reading defenses and getting the ball to the open guy, targets will probably vary a lot week to week and snap percentage might be a better predictor of success, since it indicates how involved in the offense a receiver generally is.
So here's a hard project for someone to do. Looking at game film, determine on what percentage of passing plays the ball is thrown to the first option, second option, etc. Then determine, for a specific receiver, on what percentage of his routes is he the first option, second option, etc. Combining these two should give an expected number of targets as a function of the number of passing plays run by a team, and this can be compared to a guy's actual targets to see how good of a predictor it is.