You can't just remove a relevant portion of their sample because it was bad and doesn't suit your argument, and Marner was still a better player through the final 2 years.
Marner: 0.99 P/GP, 0.95 P/GP (no EN), 0.77 P1/GP, 2.47 P/60 at 5v5 (0.58 P/GP), 2.03 P1/60 at 5v5 (0.48 P1/GP), 7.34 P/60 on the PP, 4.89 P1/60 on the PP + additional PK impacts
Rantanen: 1.10 P/GP, 1.02 P/GP (No EN), 0.79 P1/GP, 2.13 P/60 at 5v5 (0.53 P/GP), 1.40 P1/60 at 5v5 (0.35 P1/GP), 6.74 P/60 on the PP, 4.95 P1/60 on the PP
Rantanen was a worse producer at 5v5, even though he had the advantage of playing with a superior player to him.
Rantanen was a worse producer on the PP, even though he had the advantage of not being on a split unit.
Rantanen didn't PK, while Marner did bring additional PK impacts.
The only reason you think Rantanen was better is because you're incorrectly giving him credit for his team's PP time he got pumped full of. But that doesn't make somebody better.