`hero[_\-a-z]` partial selector strips article hero images and captions

[MaxWolf-01]

hero[_\-a-z] partial selector strips article hero images and captions

The partial selector hero[_\-a-z] removes any element whose class contains hero-… or hero_…, including legitimate article hero images and their captions.

Example

https://www.crosslabs.org/blog/artificial-life

<div class="blog-post-wrap">
  <h1 class="h1">Artificial Life</h1>
  …
  <div class="blog-hero-image-wrap">
    <img class="blog-hero-image" alt="…" src="…artificial_life.jpg" …/>
    <div class="blogherocaption">
      <div class="blog-image-caption">A depiction of artificial cybernetic entity… Image Credit: Generated by Olaf Witkowski using DALL-E version 2, August 21, 2024.</div>
    </div>
  </div>
  <div class="rich-text-element w-richtext">… article body …</div>
</div>

Expected: hero image + caption present in output.

Actual: both removed. _removeCoverImage doesn't apply — og:image points to a different asset.

Debug trace:

{ reason: "partial match: hero[_\\-a-z]", text: "" }   // <img class="blog-hero-image">
{ reason: "partial match: hero[_\\-a-z]", text: "" }   // <div class="blog-hero-image-wrap">
{ reason: "partial match: hero[_\\-a-z]", text: "A depiction of artificial cybernetic entity…" }

Affects any site whose content image wrapper class contains hero- (common Webflow pattern).

Fix direction

Skip the match when the element is or contains a <figure> with <figcaption>, or restrict hero[_\-a-z] to elements above the content boundary.

[mvanhorn]

Implemented in #289 - tightened the hero selector to require word-boundary anchors so 'hero-image' and similar article-hero classes stop matching.