Thanks for the feedback!
I see that you already have
opened an issue there, I'll refer to it when we start implementing.
Your solution seem more clean, and something we thought about, for us the problems were that the because changing in prometheus itself might compromise our whole monitoring infra we choose the external service discovery + multi-target exporter route to be safer.
A feedback I would give on implementing the multiple aws service discovery services on the same object, would be that their output and configurations might differ and you might end up having a complex base element.
I'm happy to collaborate on that effort if you want.
Thank you again!