Unify image alt text and caption syntax across .qmd and .md posts

Currently, authors need to remember different syntax for alt text and captions depending on whether they're writing a .qmd or .md post:

.qmd posts

Alt text uses the fig-alt attribute, caption goes in []:

![Caption](image.png){fig-alt="Alt text"}

But fig-alt renders to data-fig-alt in the hugo-md output — not to alt — so the alt text is lost for screen readers. (Tracked upstream)

.md posts

Alt text goes in [], caption uses a title string:

![Alt text](image.png "Caption")

Ideal state

Authors shouldn't need to think about which format they're in. We should find a way to make image accessibility and captions work consistently across both formats — ideally with a single syntax, or at minimum with a render hook or shortcode that normalises the output.

Compare how image syntax is rendered in .qmd and .md posts, focusing on the Hugo-md output and the reported data-fig-alt result. Investigate whether a render hook or shortcode can normalize captions and alt text, then verify that both formats produce consistent captions and accessible alt attributes for screen readers.