That chart does not show any trend of players getting better. (In fact, it would be impossible for it to show that, even if that had happened, and there is no evidence it did.)
Where there is an upward line on that chart it shows the cumulative result of a player playing better than their rating - hole after hole.
For example, Geoff Bennett's rating is just 989. Based on that, his expected final score would be 191.63. For large chunks of the tournament, he got scores on many holes that were lower than the average score for 989 rated players. So, his expected final score kept getting lower (up is better=lower score on that chart).
The reason these charts usually show upward lines (for all but Paul and Ricky) is that to get into the top ten, anyone else has to play better than their rating over the course of the tournament. Somebody in the second tier of players is going to get lucky. Normal humans need to get as lucky (or suddenly learn to play better than their rating) to beat the lucky players.
Paul and Ricky can play at their rating and get in the top ten. Paul's expected final score before he started playing was 165.7, which would have been good enough to win. So to win, Paul just needed to get scores that were expected of a 1052 rated player, which would be a flat line (with some jiggles because expected scores are fractional while actual scores are integers).
Here are some tracks of players who were highly rated, but did not play as well as expected.