So the node that got it screwed us?
"Screwed" is probably an exaggeration: that's only 3% on one block and the node lost these 3% too. That could be unintentional too (a freshly restarted node which didn't have the time to fill it's memory pool will generate a block with fewer transactions).
What's important is that it's not an exception: that's the second time I check recent P2Pool blocks and I found that they were small ones which missed potentially higher fee income.
whats the pro and cons of integrating MUCH TX in a P2Pool Block?
Larger blocks could increase the chances of losing the block as an orphan, but in the case of p2pool thanks to tx pre-forwarding this concern is mostly obviated.
The other drawback to too many tx's is GBT can take longer. Normal pools do GBT far less often than p2pool. During the recent spam txo attacks, you've seen p2pool miners increase the minimum tx fee so GBT would be much faster. 0.8.2 improved that performance issue a *lot* thanks to sipa.