I agree with what you say about pesticides -- note I said BPA is not comparable with GMO, not that BPA is not comparable with pesticide. In this case, the genetically modified organism would be safe, but chemicals used to treat the organism could be harmful.
> If for no other reason than they can produce a lot of a completely unknown and unresearched chemical.
I don't think this is true. Metabolic engineering is really hard. In other words, getting an organism to produce a new small molecule is really hard. If it weren't, we would have large scale biosynthesis of lots of stuff, instead of relying on chemical synthesis or isolation. Most enzymes have very specific functions. E.g. an enzyme will add a carbon at a particular location on a very specific substrate molecule. There's not really the possibility for unforeseen byproducts.
What you describe isn't really plausible, which is why biologists agree GMOs are safe. A plausible hypothetical for how something bad could happen, and then not be noticed until the product is fed to millions of people, probably doesn't exist (based on the fact that one has yet to be proposed).
The only exception I know is allergenicity, but this is a general risk. People are allergic to stuff. Any new product is potentially allergenic. All of a sudden quinoa is in vogue.... people could be allergic to quinoa. No unique risk.
The toxicity of _Hedysarum alpinum_ is an evolved trait, like most toxicities, that confers some benefit to the host organism. It's hard to accidentally create a beneficial new trait in organisms. It's even an academic problem in the field--evolution of novelty. Most genetic changes you make either break stuff or if you're lucky do one specific thing.