Fixes #12894, allowing the reader to find suitable preview images for posts more often.

Feed list

Site list

This PR has two parts:

1. Fixing an accidental regex breakage

A huge checkstyle refactor from two years ago ended up accidentally adding spaces to a few regex patterns used by the Reader to parse images. This almost entirely broke support for preview images that weren't set as featured. I undid the breakage in 0379e37 (and added some tests).

2. Adding support for srcset and data-large-file img tag attributes

The above fix improves things a bit but still fails to recognize images in several kinds of posts. This is especially noticeable in the site view. As explained in #12894 the /read/feed/ and /read/sites/ endpoints return different content sometimes. For some reason one of the differences is that some images have no width attribute in the /read/sites/ endpoint, so what works in the feed doesn't work sometimes in the site view.

To mitigate this, I added support for a few new img tag attributes that seem pretty common in WP posts: srcset and data-large-file. These are checked for in a fallback style.

To test:

With and without this patch, follow a few sites with a variety of posts containing images and compare.

Mainly looking for posts with a story block, and posts with image blocks (but no featured image set). (I can recommend a few.)

(For Automatticians: please proxy on your device to fully test posts with the Story block, since depending on the endpoint those won't work correctly otherwise. See pbArwn-14B-p2.)

You can also check the regex fix individually by checking out 86dedbe first and running the (failing) unit tests added.

Notes:

• Some stories (where the first slide is a video) still won't show a preview image - I'll look into that separately
• Detail view of stories is broken right now

PR submission checklist:

• I have considered adding unit tests where possible.
• I have considered adding accessibility improvements for my changes.
• I have considered if this change warrants user-facing release notes and have added them to RELEASE-NOTES.txt if necessary.