And 90%+ of the time that'll be the batsman that gives.
The bowler has the ball, the bowler controls the game. It's incredibly hard for a batsman to have a good SR if the bowler doesn't let him. And if he tries, there's a good chance he'll pay for it.
Hence, against consistently high-class one-day bowling, an average of 40 at a SR of 70 is pretty good, better than an average of 32 at a SR of 100 against poorer bowling.
I'd say, more often the not, the batsman will return back to the sheds with his high strike rate in relative tact but his average (even if low) reduced. This would, in turn, produce bowlers with higher economy rates and lower averages though - something that your all-or-nothing-economy-rate analysis doesn't really allow for.
Over the course of a bowler's career, this will generally even itself out, so I tend to agree with your career analysis of ODI bowlers more often than not. However, in isolated matches, accurate "economy" bowlers could easily end up with 10-60-4 and still have it considered as a good bowling performance, as batsmen have gone after them, got away with a few lucky shots, then been dismissed going for a ball that wasn't there to hit. This is common practice is games where a team is chasing a large score, for example, and your game-by-game analysis never seems to take in this possibility, which I have a problem with. Good, accurate bowling can still result in high economy rates for one-off games if batsmen go after it - but it will usually result in wickets as well.
So, in short, I think your career analysis of a bowler's economy holds up perfectly fine as a very accurate measure of a bowler's effectiveness, but your one-off-game analysis of a performance often doesn't really carry.