re: "when one molecule in a plant acts as a poison, then generally there will be some other molecule in the plant to counteract that poison"
How does that happen? From an evolutionary point of view, it seems a bit unlikely. Unless it's part of the plant's reproductive strategy, a plant's goal is usually not to be eaten. (However, animals often evolve to counteract plants' defenses.)
Mutation 'A' has has both benefits and drawbacks, so mutation 'B' fixes some of those drawbacks. Sure, it's good for a plant to be poisonous to animals that want to eat it, but not at the expense of poisoning itself.
How does that happen? From an evolutionary point of view, it seems a bit unlikely. Unless it's part of the plant's reproductive strategy, a plant's goal is usually not to be eaten. (However, animals often evolve to counteract plants' defenses.)