Port `SDL_LockTextureToSurface()` and `SDL_UnlockTexture()`

[yunline]

Possible implementation (pseudo code):

class Texture:
    ...
    def get_surface(self, rect: Rect):
        Surface *sf;
        SDL_LockTextureToSurface(self.tex, rect, &sf)
        return sf
    
    def update_from_surface(self):
        SDL_UnlockTexture(self.tex)
    ...

Or something better than this.

[Starbuck5]

So my thought here is why does SDL have SDL_LockTexture at all? It's enabled by the texture being set up with a flag, but couldn't a general purpose method move things from a Surface to a Texture and specifically use the fast path if possible?

I think we should not expose this implementation detail from our users.

The real question here is how should contents move between Surfaces, Textures (Images?), and Renderers.

Currently:

Texture/Image -> Renderer
    Renderer.blit
    Texture.draw

Texture/Image -> Texture/Image
    Renderer.target = xyz && Renderer.blit
    Renderer.target = xyz && Texture.draw

Surface -> Texture/Image
    Texture.update

Surface -> Renderer
    No path

But what if you could just blit to things and have the internals figured out by us?

Texture/Image -> Renderer
    Renderer.blit
    Texture.draw

Texture/Image -> Texture/Image
    Texture.blit

Surface -> Texture/Image
    Texture.blit

Surface -> Renderer
    Renderer.blit