Respect viewport position in coordinate conversion functions

[Novakasa]

Objective

• In Camera::viewport_to_world_2d, Camera::viewport_to_world, Camera::world_to_viewport and Camera::world_to_viewport_with_depth, the results were incorrect when the Camera::viewport field was configured with a viewport position that was non-zero. This PR attempts to correct that.
• Fixes Camera::viewport_to_world_2d doesn't take into account the viewport position #16200

Solution

• This PR now takes the viewport position into account in the functions mentioned above.
• Extended 2d_viewport_to_world example to test the functions with a dynamic viewport position and size, camera positions and zoom levels. It is probably worth discussing whether to change the example, add a new one or just completely skip touching the examples.

Testing

Used the modified example to test the functions with dynamic camera transform as well as dynamic viewport size and position.

[alice-i-cecile]

Migration Guide

Not sure if this does this need a migration guide entry, since this can probably be considered a bugfix.

This PR changes the result of the functions Camera::viewport_to_world_2d, Camera::viewport_to_world, Camera::world_to_viewport and Camera::world_to_viewport_with_depth in the case when the viewport position is not UVec2::ZERO. With a viewport position of UVec2::ZERO, the behaviour is not changed.

Good migration guide, but yes, bug fix :)

[tychedelia]

Thanks for taking the time to update the example to make sure this doesn't regress. The example looks a little funny on macOS, likely due to window scale factor, although seems to be functional. I'd suggest testing with scale factor = 2.0 to emulate macOS.

[Novakasa]

Polished the example quite a bit

• Initialize viewport relative to physical window size (should hopefully fix MacOS weirdness)
• Controls now in FixedUpdate and respect fixed delta time
• Simplified viewport bounds visualization with a background mesh
• Draw cursor after Transform propagation to get rid of camera control lag
• Add comments

[Novakasa]

@tychedelia would appreciate another test of the example, since I think the scale_override behaviour on Windows might be bit different than MacOS defaults.

[tychedelia]

Looks great and even more improved! Thanks.