Consider removing the `ImageDecoderRect` trait
The ImageDecoderRect trait is supposed to help decode only the specified part of the image, as opposed to the entire thing.
As of v0.25, only BMP and Farbfeld decoders implement it: https://docs.rs/image/0.25.4/image/trait.ImageDecoderRect.html
Given the lack of format support, it's probably best to remove it for now, to reduce the API surface and maintenance burden. We could possibly re-introduce it later if there is enough demand.
FWIW imagemagick does provide this functionality, so a proper implementation of this could potentially be of use to wondermagick. But it is only an optimization, ideally and we'd have to generalize it to animations as well, so I don't think it's worth the trouble at this stage.
I think this is worth considering. Part of the problem with ImageDecoderRect is there's formats were it is sometimes possible to extract portions of the image, but it depends on the choices made by the encoder.
If we do decide to remove, thing thing to do would be marking the trait as deprecated for a few releases and see if any concerns are raised.