Matt talks about performance within regions (recognizing that networks and sites in different parts of the world, eg, have different capabilites), and indicates that, because of these differences, Google now tends to look at "outliers", performers significantly below the norm. So, right now, "really, really slow" sites do suffer in rankings... pretty much what I'd noted above. Matt says that while the question was about mobile, the current speed signals apply across the board to all sites.
If you listen to the answer closely, Matt suggests that Google is also paying attention to certain performance factors that might particularly affect the mobile experience. If the effects become significant enough, Google may, over time, start applying those mobile speed signals to ranking.
So, Google clearly is paying attention to quality signals over time, which leads into your second question, about the Amit Singhal blog post from early April, 2010...
...the blog post talks about the introduction of a "new signal". It makes me wonder if it became an indirect contributing factor to Panda's QA considerations.
The post was published several weeks before the MayDay update around the first of May, 2010. MayDay, IMO, was precursor to Panda... and yes, I'm sure that site speed was one of the early quality signals that Google's seed sets were evaluating for what I see as a continuum of updates. At this point, I'm not sure exactly what was blended in when... but all of these factors continue to be signals that Google pays attention to.