Hallo @Matthias,
Thank you for spending some time on this.
What I have learnt so far, is that the preferred way seems to be working with Pages instead of Tables - the reason is structural, there seems to be bit less messy content.
I agree that it might be easier to decompose it into “easy” and “hard” part
I even mentioned that earlier.
I cannot figure out the appropriate sequence of steps, meaning whether to deal with the offset first, or maybe filtering. There is a post which I find quite inspiring :
It seems that the kind of problem is similar, yet the logic of solution provided by @Melissa is pretty advanced. I am first to admit that it looks elegant and efficient, I will try to use it in this one - need to study it further though.
Your file is on (mostly) and off (sometimes), the guidance as to whats relevant and the desired structure is attached in my first post.
Once again thanks for your input 