It isn't really 100% of 0.6%... copyright isn't based on byte-count. You may or may not be able to get a straight answer about what it is based on out of the literature, but it's certainly much closer to the human experience rather than the raw byte count. In legal terms, it's basically 100% of 100% of the article being republished. (I'd say it's 100% of 100% if the header image had been included.) Nobody showed up to that page for the 4MB of other stuff, minus the aforementioned header image. Strengthening the case for infringement is that monetary damages are also being experienced, because what is mostly being stripped is the revenue generation portion of the performance of the work.

Basically, if you take the four criteria for something being fair use [1] in the US, which isn't the end-all, be-all of analysis but is a darned good start, this absolutely faceplants on all four. (And while it is technically up to the discretion of the judge, I'd say it's closer to an "AND" than an "OR" on the 4 criteria. You really, really don't want to be standing in front of a judge solid on only one out of four. This is to preclude the usual arguing about #1; even if I grant it, which I don't, it really isn't enough on its own.)

[1]: https://en.wikipedia.org/wiki/Fair_use