I wouldn't be surprised if the actual death toll is somewhere in the high hundreds. thousands seems maybe high, as it was at night so there would be fewer workers in the immediate blast radius. but I'm not entirely familiar with the population density in urban areas in Beijing.
Well, my worst fear is that the first explosion and the fire attracted people to the windows - and then the big one went off. It's pretty much as bad as it can get, maximizing the amount of people near windows just before the shockwave. Expect a lot of people that moved to the side of building from which the fire was visible - these people moved from safe side to the exposed side before being hit by the blastwave, shattered glass and in some cases by initially supersonic debris. Did you guys have a look at some of those fiery debris ejected by the blast? By the time the initial flash dissipates there are already a lot of fiery arcs that have already slowed down considerably.
Google says 5,400 per square kilometer. The blast radius was at least a kilometer... so 2 kilometers total diameter.
I think it's safe to assume that the density in that particular area was a little lower, but yes, generally there abouts.
Well, tricky. On one hand it's harbor which reduces density but also it's harbor - which means that some of the tallest and 'densest' buildings are located in the immediate vicinity whereas the pop density usually drops the further you move away from the harbor and shoreline.
If that wrecked house at the edge is 5 stories and 1 story is ~3m then the crater is around 70m in diameter, looking at it quickly by finger measure.