Is there any possibility for foul play here? I.E. the top 6-10 pools have agreements that...
No. As you can see at the same link, 466705 by Kano pool on 2017‑05‑16 is a full 40 seconds later than Bixin propagated to blockchain.info but we won the orphan race regardless. This can be possible due to an unfavorable circumstance within a single BGP ASN area. Imagine a broken sea cable, and a small backup line which gets fully congested and lacks throughput during peak hours. Furthermore, complete outages of 15-30 seconds between two big ASNs always occur, a simple BGP routing issue somewhere in the Internet does suffice. No broken cables or devices required.
The catch there is that the times shown by many sites is the timestamp in the block header.
That timestamp is not when the block was found, or even when the web site first saw it.
It's up to the pool to decide what to put there, then it sends the work to the miner, who may later find a block, and here that will be some time in the 30 seconds after the work was sent to the miner, with an already 'locked in' timestamp.