5v5 SV% is just a more reliable and repeatable predictor of performance. Larger sample-size and less systems-dependent.
On a game by game basis, it doesn't really matter. If a goalie stands on his head during the PK, good for him.
It's just not something you should expect the next game. And if a goalie has a great 4v5 SV% one season, you shouldn't take it as evidence he's somehow particularly great at shorthanded goaltending, because it's probably not going to happen again next season.
For a team like the Blackhawks, who are rebuilding, it doesn't really matter how much of what Lankinen is doing now is repeatable. We're not dealing with anything resembling a finished product as far as the play in front of him, not to mention the PK system and talent is liable to be completely different season-over-season as kids grow and new players are acquired and others are shipped out.
It's now the same as a team like the 2013-2015 Blackhawks, where the primary players, the coaching systems were all fairly static year to year. Then you could and should actually look at the predictive numbers for both the goalies and the players, to see how well they're actually playing versus how lucky they're getting.