You have to remember the fiber is extremely small so even bendy light approximating the curvature of the Earth wouldn't affect it. The thing that would affect it is a variable angle of reflection that a cable following the curvature of the Earth would create.
Fixed.
Or rather, the two would be equivalent.
I'm sorry mate, you are way off here. Fibre optical physics relies massively on straight light.
The curves cited wouldn't produce your diagrams - quite simply, if light bended that much on a regular occurence, what you and I see as a straight door frame, from a physical matter perspective, wouldn't be straight at all. Light bounces many millions of times down a length of fibre optic - and we are suggesting deviation (completely hypothetically as no serious values have been decided on to my knowledge? I will happily stand corrected here) of around a micrometre per metre or 2 of distance travelled. On that basis, it probably would take a fairly large distance, a few miles maybe, to properly see the effects described. Since we know that fibre optic cabling is used for many thousands of miles around the earth to carry our communications - we can deduce light surely cannot bend even on a really ridiculously small level.
I think there are a couple of people here that really need to read up on fibre optics - including incidence angles, critical angles of a fibre optic boundary, total internal reflection and a culmination of all these topics which is the acceptance cone of a fibre optic. Most importantly, bending light would violate the concept of an acceptance cone INSIDE the fibre optic; which anyone with an ounce of knowledge on fibre optics, know cannot be done. And do y'all know why? Light is straight peoples...that's the end of it.