I haven’t run Apocalypse World either, but I’ve used that technique in other games (started a fantasy campaign using Open Legend that way). It’s not a bad technique, but it’s not universal. The type of conflicts it generates tend to be personal, which might not be appropriate for a campaign where everyone expects a call to action to lead to a plot where the PCs have important parts to play.I've never GMed Apocalypse World, but I think it has an interesting set-up for the first session: rather than a GM hook for an adventure, we learn about the PCs in their world, and rely on the rules for narrating consequences, especially failures, and for framing (ie soft moves followed by hard moves) to make things happen.
I think complete aversion to metagaming is pathological. Even if there are obvious mechanics at play (such as in your Traveler scenario), the players are arguably going through the same decision-making process their characters are: jobs don’t come often, there aren’t many patrons we know, etc. I think it would be better to accept that some metagaming is diegetic (and actually good) than to try to eliminate it completely.The "metagame" dimension comes in via the rules for framing and for consequence narration, which direct the GM to do these things in a particular sort of non-"neutral" way.
We had a situation in my campaign where this pathological aversion reared its head. Even though the campaign is about what the players want to do (see: my comment in another thread on the structure I use), they got worried they were “metagaming” because they had found what they thought they were looking for, but they had not found the key yet. There were several keys that would open it, but they had a thief. He could also open it. They were worried that having the thief pick the lock was “metagaming”.