I believe this statement is fundamentally incorrect. I don't think 'eye witness' testimony has ever been thought to be 100% reliable. Ever.
I think perhaps you may be misunderstanding Schraf. It seems to me that she didn't say that eyewitness testimony was always considered perfectly truthful and accurate, but rather, there was the presumption that even if an eyewitness turned out to be lying, they did still have an accurate memory of what they did see, even if they chose to testify a lie, instead.
The idea that someone could honestly, actually remember events that did not happen is, I think, rather new. Though the idea that one might not remember what one saw is not. In other words what is new is the idea that an eyewitness could give honest testimony that didn't actually happen. The idea that memory "failures" might be additive as well as subtractive in a completely healthy sober adult is what is new, I think.
Or maybe it's not. I'm not familiar with the history of legal argumentation.