For Millsaps, I had to remove San Diego 1, because as you can see in the by-race graph, Millsaps had one early lap that throws everything else off. Once SD1 was removed, he averaged 0.031 seconds increased per lap -- in a similar range with Dungey, Roczen, and Tomac.
Looks like maybe something was up with A1, as well, for Millsaps. While I don't really love the idea of just removing data if it looks odd, if we did remove it, it only increases his average increase per lap to 0.047 seconds -- still not particularly high. Here I think we're seeing the problem with only using 5 races as the data set, but that's on me ;)
Seely has had quite a bit of fluctuation from lap to lap over these first 5 races. Maybe that's just noise from not enough data, and it will flatten out as the season goes along. His average increase in lap time is 0.074 seconds, which is the highest individual average we've seen so far, though not by a particularly large amount.
Seely's odd race is A2, where he had two laps at the beginning of the race that were much, much slower than the rest of his laps. If we remove that (again, not that we should), his average increase goes up to 0.084 -- so, it goes up some but not by a whole lot. There's a similar problem with SD1, though not as pronounced; if we remove that, his average goes up to 0.131, which is extremely high, but again is now relying on very limited data.
The last rider suggested, Mike Alessi, differs from the others in that he's typically riding in the bottom-half of the pack, rather than the top-half as we would expect from the others. That might change how we view the WHY of the increase lap times, but his average is substantially higher than the others we've looked at: 0.096 seconds increase per lap on average.
As you can see, Alessi had a "decrease" in seconds per lap in SD1, which looks attributable to a problem of some sort on Lap 8. If decided that SD1 was not indicative of how he normally rides, the average increase without that race would be 0.127 seconds per lap.
I'd like to add in Pourcel because I (like many others, I think) am mystified by his amazing ability to put up a great qualifying time in the practice laps but then he tends to disappear in the races. I've heard two explanations that make sense but are difficult to verify: he hasn't built up the endurance because he's relatively new to the 20-lap Supercross races, and/or he isn't comfortable with contact once the rubbin's-racin' begins. If it's the former, and he's fading due to fatigue, we would see a big increase in lap times as the race goes along. If it's the latter, then I think we'd see relatively slower laps at the beginning of the races, before the pack has thinned out. If it's some combination of both, then I'm not sure what we'd expect to see. And of course it could be something else entirely.
For Pourcel, I'm going to remove A2 because he had a problem in Lap 3, which leaves him with a negative result for the race as a whole. After doing so, his average overall is 0.100 seconds increased per lap.
You could also make a case to remove Oakland because he had what must have been a crash in Lap 17 -- removing that result brings his average to 0.071. So, Pourcel does seem to lose more time per lap than the other riders who are generally putting up the fastest practice laps (Dungey, Roczen, Tomac, etc.), which may speak to Pourcel not having a level of riding that can carry through to the second half of the race, perhaps as has been suggested he has not yet built up his endurance.
As for the riding in traffic issue, I think we see something for that, too. If you look at laps 2 & 3 for Pourcel, they are much higher than laps 4 through 14 where he seems to have a pretty consistent pace (albeit increasing slightly, as we would expect). If you compare that to the previous 3 riders we looked at, their laps 2 & 3 may be higher, but the dip is not nearly as pronounced:
It seems clear that -- for whatever reason -- Pourcel is going much slower in laps 2 & 3 than in laps 4 through 14, compared to Millsaps, Seely, and Alessi. So it looks like Pourcel is having trouble keeping his speed throughout the race (there's a noticeable spike after Lap 14), AND his early laps are slower than his laps in the middle of the races. Hopefully that provides some insight into what's going on, even if it doesn't completely answer the "why" part of the question.