Fixing Y12i stream output in MIPI with CUDA case

[Arun-Prasad-V]

Tracked by LRS-1052

[Arun-Prasad-V]

The code for mipi and not pipi is exactly the same - the only difference is the padding of 8 bits that has been added for mipi. Few questions: Is the other padding used anywhere? (y16_y16_from_y12i)? If no - please remove it, and keep only the mipi one. If yes - since the code is exactly the same besides the type used (y12i_pixel or y12i_pixel_mipi), could you change it to a template function that would be used for both cases (once with the y12i_pixel type, and once with the y12i_pixel_mipi type)?

@remibettan, UVC with CUDA must be using y16_y16_from_y12i. So, as you mentioned, will change them to template functions.