The problem with using league average numbers is that, even if you believe defense doesn't have much of an impact on save percentage, PPs undoubtedly do. Smith has faced the most PP shots against by over 100 over his tenure with the Coyotes.
That's an interesting point. Has anyone crunched the numbers to see how those extra PP shots impacted his save%?
What striking to me about that statistic is that he's only 6th in minutes played during that timeframe. Which is still a lot, but hints at his taking PP shots at a very high rate relative to his minutes in net.
You don't need to be in the 25 best to be above average. There are also picks made based on chemistry and perceived team needs. They don't just randomly pick worse players.
But you are clearly acknowledging here, that there is more to roster selection than how good a player is. There's not a chance in hell that Kunitz makes Team Canada based on being a better player than the guys he beat out.
To be specific about Smith, his selection for the Canadian roster had a lot to do with 1) having had a strong playoffs in 2012 and getting that "tournament goalie" reputation, 2) playing for Team Canada in the World Championships the previous year, while other top goalies were preoccupied in the Stanley Cup playoffs, and 3) being experienced and well suited for the large-ice style of play.
Here's an InGoal article on his selection -- note that it leads with a couple of comments on how questionable his selection seemed at the time:
http://ingoalmag.com/news/mike-smith-canada-makes-sense-sochi/
Because let's be honest, there's no way that 2014 Mike Smith is a better overall goalie than Price, Luongo, Holtby, Crawford, or even the 2014 back-from-the-dead version of Mason or fluke-mode Bernier. That's pretty obviously more than 3 Canadian goalies who could have made the roster over him, and further confirmation that if he's ~5th among Canadian goalies then he is de facto an average NHL starter.