"Aside from performance optimizations, such evaluation of the 103 (Early Hints) response's header fields MUST NOT affect how the final response is processed."
But beside, if you are in a position to insert HTTP responses, you are in a position to inject arbitrary JS.
Really? Let's see, at least one extra DNS lookup can happen, a URL fetch can happen, this is totally enough for advertising tracking.
It's a sneaky unwanted network request un-controllable by end users. And it happens even before you load the page.
All ad-blocking tools/addons today works by parsing the webpage content, the 103 header extended the game from the body to the header.
Advertising tracking at its best.
[1] https://w3c.github.io/preload/
[2] https://bugs.chromium.org/p/chromium/issues/detail?id=471199
I think this is the case where X is innocent and Y is innocent, but combined X and Y leaves a huge exploitable surface.
No, no they don't. They primarily work by hooking network requests through e.g. chrome.webRequest[0] and blocking certain URLs from being loaded. This continues to work with the multiple header-based mechanisms for downloading things.
So yes, the file would get fetched, but no, it can't affect the page being rendered by the browser.