Further, if you directly view an image and it breaks half way, the browser will hide it and say it's corrupt (and have no "show anyway" button), unless the size wasn't provided, then the browser just thinks it's valid, despite that it could parse the JPEG file and find out that lines of pixels are missing. Calling either of these cases a feature seems very biased.
We aren't even talking about one of the worse problems, which is when a page fails to load, the browser doesn't try to reload it - you have to come back and press reload yourself. I've wasted literally hundreds of hours doing this.