ndc_to_world seems to have wrong output range in documentation

[jwir3]

How can Bevy's documentation be improved?

Inside of bevy_render/src/camera/camera.rs, the following documentation for ndc_to_world states:

    /// Given a position in Normalized Device Coordinates,
    /// use the camera's viewport to compute the world space position.
    ///
    /// When the position is within the viewport the values returned will be between -1.0 and 1.0 on the X and Y axes,
    /// and between 0.0 and 1.0 on the Z axis.
    /// To get the world space coordinates with the viewport position, you should use
    /// [`world_to_viewport`](Self::world_to_viewport).
    ///
    /// Returns `None` if the `camera_transform`, the `world_position`, or the projection matrix defined by [`CameraProjection`] contain `NAN`.
    ///
    /// # Panics
    ///
    /// Will panic if the projection matrix is invalid (has a determinant of 0) and `glam_assert` is enabled.

I believe this should state:

    /// When the position is within the viewport the values returned will be between `-world_max_width/2.0` and `world_max_width/2.0` on the X axis, `-world_max_height/2.0` and `world_max_height/2.0` on the Y axis, and between `0.0` and `world_max_depth` on the Z axis.
    /// where `world_max_width` is the maximum world width, `world_max_height` is the maximum world width, and `world_max_depth`
    /// is the maximum depth, in units (by default, pixels) defined by the user. 

For reference, when running the above code in my world, I can definitely get it to be outside the -1.0 to 1.0 bounds. It seems like this is an input constraint, not an output constraint, though I wanted to be sure before I posted a PR.

[jwir3]

For some additional context, ndc_to_world() seems like it should take, as input normalized device coordinates (-1.0 to 1.0) and output world coordinates (which, by default, for a 2D scene, should be pixels, correct?)