There seems to be something to add to your reference that was left out for good or bad.
The mercury emission from a coal fired electrical plant is distributed over the earth in the atmosphere and the mercury emission from the bulb is concentrated at the place where it is disposed of and then this concentrate gets into the soil as a direct contaminant. Assuming recycling, even that is lossy and some mercury would probably get into the environment.
What is worse, 2 mg of mercury in a small stream running out of a waste site or 4 mg of mercury being emitted from the stack of a power plant and dispersed over a whole city? The amount per unit area is actually lower in the scond case, I suspect.
IDK which is worse, but I assume that they are both equally bad in spite of one being neary 2x greater than the other. The problem is, no one really knows for sure and that is part of the whole problem. Some say NO mercury should be placed into the environment; that it should be a closed loop with no loss at all or should not be used.
This is the problem. And, considering the amount of coal used for lighting vs coal used for running other equipment, lighting is only a part of the problem, not all of it. If everyone converted to fluorsecent lights and then lighting use was included in overall power consumption, we might not even be able to detect the difference.
People just jump quickly to a conclusion. 'This is better than that', with a 'statistic' to show, but it has been said that anything can be proven with statistics.
Mercury is bad. We should eliminate its use or escape into the environment. The fluorescent bulb is not all of the answer and may not even be the best one.