display: fb: use a single plane to describe the packed RGB frame buffer

No need to allocate a picture for that. It was not allocating any data, we forced the one plane values anyway.

We can use plane_CopyPixels() directly on that single plane.

When we know the bits_per_pixel we request the core the matching packed RGB format with the same width/depth.

If the forced chroma is not a packed/single plane chroma the plane copy won't be sufficient, but that's a pre-existing issue. Only one plane is currently supported.

We could also use a real picture_t with the actual RGB chroma. But since we force one plane dimension it has no benefit, just more overhead. In addition the current picture visible plane values were kept from the original vd->source chroma+dimensions which can be very different than the actual plane dimensions we have.