[Bug] Generative fill fails when input image is larger than 25 megapixels

[tpiros]

Bug Report

Describe the bug

Using generative fill with images that are bigger than 25 megapixels is causing a transformation error. Adding a c_limit before a b_gen_fill call in the transformation chain gets rid of this issue.

Is this a regression?

No

Steps To Reproduce the error

1. Create a project with one of the community SDKs
   eg. Svelte:

<script>
	
  import { CldImage } from 'svelte-cloudinary';


</script>

    <CldImage
  src="demo/nice-red-shoe.jpg"
  width="960"
  height="600"
  alt=""
  crop="pad"
  fillBackground
/>

Add an image where the MP is higher than 25MP. Play around with fillBackground and crop options.

Expected behaviour

fillBackground should restrict the image to match the MP limit

CodeSandbox or Live Example of Bug

https://codesandbox.io/p/devbox/mqsvyn?file=%2Fsrc%2Froutes%2F%2Bpage.svelte%3A5%2C39

[eportis-cloudinary]

This affects more than just generative fill, it affects any transformation where we resize in a second step (if the input image, or any intermediate artifact after e.g. padding, is >25MP).

For instance:

https://res.cloudinary.com/eric-cloudinary/image/upload/c_pad,ar_1:1/w_1000/v1/24mp-image.jpg

gives a 400 Bad Request with X-Cld-Error: Maximum image size is 25 Megapixels. Requested 25.01 Megapixels.

However, https://res.cloudinary.com/eric-cloudinary/image/upload/c_scale,ar_1:1,w_1000/v1/96mp-image.jpg does not trigger the error.

Instead of only fixing this for generative fill I would like to insert the c_limit any time we are transforming in multiple steps, and an intermediate step could possibly produce a >25MP artifact. Will need to think about how to do that cleanly, either in all of the relevant plugins or somehow across all plugins. This will take some careful consideration.