vout_subpicture: separate region picture and region cropping

Each region contained a video_format_t and a picture with its own video_format_t, which any clear indication of which is supposed to be used for what.

Now the palette is always used with the picture video_format_t and not the side video_format_t, which is only used for positioning the picture. Only the x/y offset, visible width/height and SAR are usable from that video_format_t.

On top of !4888 (merged). Extracted from !4905 (closed).