Include smaller covers in images along with image dimensions

Right now we always include and load the largest image.

We may be able to make the loading a bit snappier by using medium size images as covers too.

See here for how they should look like
https://specs.opds.io/opds-2.0#23-images

Notable, we need to include image dimensions if that is possible in order to be able to tell which one is small and large.

Including the image dimensions would also allow us to better shape the skeleton loaders.

Start by locating the Open Library cover-handling entry point and the code that currently selects and loads the largest image. Read the OPDS 2.0 images specification linked in the issue, then trace any existing tests for cover data. Done means medium and larger covers can be represented with their dimensions, and the available dimensions support better skeleton-loader sizing.

Written by the indexing model from the issue text.