You might wonder how I got inspired to alter my prism design so I could elongate the scanline.  I was reading up on military applications for laser scanning and found an article on lidar scanning.

I found the following patent application US20190310468A1.  The patent is abonded and irvine sensors seems bankrupt.
One of the figures is provided below,

Let's first look into the problem they are trying to solve. They try to improve the duty cycle of lidar. A laser bundle has a finite size and gets clipped at the edges. 
So by alternating between two reflective polygons they can fix it. This is key in certain mission-critical LIDAR applications such as airborne collision avoidance, military reconnaissance or autonomous automotive applications.

This system also has a sort of rotating input mirror for incoming beams. These get reflected and redirected to two reflective polygon mirrors, which are optimized for the outgoing lasers.
The system seems to me hard to align and make.

Anyhow, this inspired me to think in terms of  input and output prisms. For prisms the outside redirection is not needed. This simplifies it greatly.

You could do it internally.

I will add some ideas for lidar at a later point