It's similar to the implausibility of group selection -- doing things for (unrelated) group members doesn't give a differential advantage for the doer's genes, so it tends not to happen. Likewise, techniques that hook you on "sugar-fat foods" ("superstimuli") in general benefit all companies in the market, not just yours.
So I don't think the vendor lock-in model has the same dynamic or is representative of the incentive set that food companies have.
[1] Pushers of illegal drugs have yet another dynamic at play, in which case finding a steady source of the drug is hard, so hooking one user will likely mean they buy from that pusher, not from dealers in general, due to the difficulties of navigating an illegal market.
Food and willpower have a tricky relationship, because willpower is literally fueled by food (glucose). When you're well-fed, your willpower is high, but when your glucose is depleted, suppressing the instinct to reach for that donut is a non-trivial proposition.
I think it's absurd to hold super-stimuli providers legally or morally responsible for poor nutrition habits; they're giving people what they want. But I do think it's helpful for consumers to view Doritos and Coke in the same category as other risky behaviors with high addiction potential: cigarettes, gambling, etc.
That's a good point, and it would be interesting to learn what keeps consumers from searching for substitutes. I guess that marketing and branding makes people so strongly associate the good feeling with one very specific brand, which makes them not want to seek out all a low-bidder.
Otherwise, I think we're in agreement, and we have to recognize that the effects of the foods make people deviate from the "level-headed decision maker" that justifies leaving people to their own devices.
I think there's a big difference from standard vendor lock-in, in that many different companies provide similar junk food.