Introduced crop parameter in image for selective display of image portions

[varphone]

Introduced crop parameter in image for selective display of image portions

This PR introduces a new crop parameter in the image function.
This allows users to selectively display portions of an image, which can be particularly useful when working with large images or when only a specific part of an image is relevant.

• Add a new crop parameter in the image and decode function.
• Support raster images.
• Support vector images.
• Support (left, top, right, bottom) tuple as the value for the crop parameter.
• Support (left, top, width, height) tuple as the value for the crop parameter.
• Support ratio based cropping.
  • crop: (left: 0.1, top: 0.1, right: 0.9, bottom: 0.9) will crop the image to 10% from the left, 10% from the top, 10% from the right, and 10% from the bottom.
• Support pixel based cropping.
  • crop: (left: 100, top: 100, right: 200, bottom: 200) will crop the image to 100 pixels from the left, 100 pixels from the top, 100 pixels from the right, and 100 pixels from the bottom.
• Support ratio and pixel based cropping together.
  • When the left or top >= 1.0, it will be considered as pixel based cropping.
  • When the right or bottom > 1.0, it will be considered as pixel based cropping.
  • crop: (left: 0.1, top: 100, right: 0.9, bottom: 200) will crop the image to 10% from the left, 100 pixels from the top, 10% from the right, and 200 pixels from the bottom.

[johannes-wolf]

I think, ratio based cropping should use Typst ratios: 10% instead of 0.1.

[tingerrr]

I wonder if this should be image specific, or instead be a more general, improved clipping + place/align mechanism on containers (block/box).

It reminded me of #2709 where I attempted to change which part of an image was shown using block(clip: true) combined with align and move.

[laurmaedje]

I'm also not sure if having this as an image-specific thing is great. The only thing that makes only sense in the context of images is the pixel-based clipping, but how frequently is that needed? I'm also concerned about the re-encoding. It's less efficient than path-based clipping and it's also easy to lose color-space information etc.