The standard setting is interesting, and usually causes the fail rate to spike up a bit. E.g. they did a new standard setting in 2008, and the fail rate jumped from 3.5% (in 2007) to 7.4% (in 2008). They did it again at the end of 2016 and it caused a smaller spike (3.4% in 2015 to 5.2% in 2016). Since the standard setting was done at the end of 2016 I'd expect 2017's fail rate to be a little higher too.
I think the image you have relates to a hypothetical standard setting - the fail rate in 2013 was still 6.3%, but had they implemented the new standard (which they ended up doing in 2016) to the same results, 10.1% of people would have failed instead.
I suspect there won't be another standard setting to Part I or Part II though, as the new combined exam is taking over.