On a file page (or diff page of file page), like this one, the “structured data” heading is a level-1 heading (<h1> in HTML, = heading = in Wikitext) instead of a level-2 one (<h2> / == heading ==). Since all the other headings below this one are level-2 (summary, licensing, file history, etc.), this gives the impression that they all belong to the “structured data” section, which is not the case.
No, that doesn't fix it, that proposes making things worse by hijacking generic code for our purposes. The slots are intentionally separated by H1s, with the main slot represented implicitly by the title H1, but our moving things around confuses things.
For the sake of visual consistency, if a structured data section has to be in a tab with other content, then all the headers, wether about structured data or not, should all be at same size. All in h1 or all in h2, no matter for me.
As I understand it, the tabs (which will replace the H1s) are "File information" (captions and the wikitext block) and "Structured data" (for all the statements and the EXIF-extracted data). Possibly we could look at having other tabs for the file history (currently shown as if it's part of the content block) and the file usage information (ditto)? Of course, all of that is subject to change based on feedback.