What would be best really depends on what you are trying to cool.
There is a lot of things to take into consideration here. CFM of the fans, how well the canopy seals and the air volume of the canopy itself. But here is a few ways to look at it.
Assuming reasonable air sealing and the "usual" amount of air space.
First option -- 2 pushing fans. 2x the airflow, and some air escapes through leaks. The air pressure would be slightly higher in the canopy, and thus you would have a higher heat transfer to the air. The fans blowing across the water or not, don't matter all that much if the volume of air is high. it is a safe assumption that the exiting air would be equal in temp to the water, and thus no more cooling could occur.
Interesting Enough, if the air volume is very high and the CFM low, you would acutally increase the heat transfer from the lamps to the water by the high air denisty and circulation.
This option also works well because you can put an air filter in front of the fans and reduce dust buildup, which insulates very well.
Second option -- 1 push , 1 pull. 1x the airflow, and equalized air pressure with outside air. Not, really any great bonus to this setup except for more control over the entry and exit points of the air (IE, no dust or dirt buildup around air leaks).
Third option. -- 2 pulling, 2x the airflow. The option works better if you are just trying to cool the water. The air pressure decreases, which would normally lower the heat transfer from the lights to the water (think a wannabe thermos). However, this option also increases evaportation of the water (things boil and evaporate easier in low peressure environments). So, you would cause your light to run a bit hotter, and transfer less of that to the water, and evaporate more water out. But only if the canopy seals well enough and has a small enough air space to acutally lower the pressure enough. (which I have not run the numbers on).
Of course, with the fans pulling, you do wear them out quickly. It's not so much an issue for me, as I have a perpetual supply of computer fans and 12v power supplies. But for anyone paying for fan, that might be an issue.
If you want to get real technical, it's the evaporation that accounts for most the the tank cooling effect, rather than the acutal temperature difference of the water. Very much so with the 0% humidity of a kamloops summer.
Now, If I could just find a cheap supply of 30cm 12vdc fans I would be in heaven. Finally a quiet fan
