As for the last item, it's not as daunting as it seems. Range safety is actually fairly comparable in either direction. The trajectory of the booster won't naturally head over populated areas and it will have to be adjusted until it reaches back to land. If, for whatever reason, the trajectory adjustment goes to far and the booster would end up landing farther back on land then the stage can be signaled to be destructed, which will screw up its drag coefficient enough so that the debris will just fall safely into the water. And that's a worst case scenario.
Another reason it's less daunting is that, by the time of the landing burn, there's very little fuel remaining in the rocket. On the way up, the rocket has to have enough fuel to power the 9 first-stage engines for 181 seconds, and the second stage engine for 412 seconds. (As well as fuel for the various return and landing burns.) So it will have over 2000 engine-seconds of fuel.
In the landing burn, it's powering 1 engine for maybe 10 seconds. So that's less than half a percent of the initial fuel mass - something that would make a potential failure a lot less dangerous.
Ten seconds seems like an incredibly liberal estimate. You need a pretty decent "margin of error" reserve. I'd wager we're looking at something more like 30-60s of fuel during the last stages of landing, at least. So let's call it something like 2-3% of the fuel.
SpaceX claims that Falcon-9 first stage has a pretty good mass ratio - meaning that the fuel weights many times more than the tanks and engines (and everything else).
However, even if we assume 30 as the ratio of loaded vs. dry mass, which is better, for example, than Titan II first stage, which has pretty good mass ratio, we need to add the mass of landing legs - something unique for the Falcon. Suppose legs halve the mass ratio, making it 15.
9 engines are capable of lifting the dry first stage, the fuel in it, and the second stage plus payload. That means 9 engines can lift much more than 15 times the dry mass of the first stage. Dividing the thrust by 9 makes the thrust much more than necessary to support the dry stage.
In other words, even the thrust of a single Merlin engine is more than enough to brake the dry weight of the first stage. That means when first stage lands the engine is throttled - spending less mass of fuel per second than it does when Falcon lifts off. The better mass ratio of the first stage, the smaller flow of fuel is needed to decelerate it - so less fuel is needed to brake. It's not enough to count seconds of thrust - how big the thrust is should also be taken into account.
The engine can only throttle down to about 70%. For the final landing, the first stage will only use the center engine (leaving the other 8 shut down).
Even with only one engine at 70%, the thrust to weight ratio is greater than 1. That means the rocket can't hover, they just have to time the burn perfectly so the speed reaches 0 at the moment it reaches the ground.
Most rocket engines don't throttle down well. The main reason isn't the turbopumps (although that can also be an issue) but combustion instabilities in the chamber and flow separation in the nozzle.
* The 3-engine deorbit burn worked; previously it was screwed up by uncontrolled roll preventing the engines from getting fuel
* The 1-engine landing burn, previously separately demonstrated by the Grasshopper and F9R test vehicles, worked for a rocket coming down from orbit
The remaining items:
* 3-engine burn to take the rocket back to the Cape from downrange
* Making everything safe enough to dare landing at the Cape, without killing anyone or destroying all the expensive ground equipment that's nearby.