In the turbo applications I am familiar with, both a wire or film (better) type MAF is used, in conjunction with a MAP, IAT, ECT, etc.
Specifically on turbosupercharged Subarus, there's knowledge to be gained by also knowing intake pressure.
Ultimately, you are trying to reach the best AFR. Things change a bit when supercharging, whether mechanical or turbo.
Things also change with humidity - a hot wire gets cooled more with higher humidity, which doesn't represent denser air. Hot film type MAFs compensate for this somewhat. But you would still like to know manifold pressure after a turbo.
I certainly agree it's more complicated: With ROMRAIDER, I have seen there are dozens and dozens of "modifier" tables in a Subaru tune that ultimately affect injection and timing on top of the MAF table. Tuning is complex with so many vectors and map layers, it's difficult to get your head around.
However, I would have to think that this added complexity serves some benefit, especially when controlling engine operation under high levels of boost. While I know there are sophisticated speed-density systems out there, cars don't run on open loop anymore (at least not for long) and the more sensor information available, the better.
I think of it this way (it's what I tell my students): Air "flow" tells you one thing (like the vane meters) MAF tells you a bit more, and MAF+MAP tells you almost everything. It's just a simulation; a device to approximate.
In fact, the wire or film assumes a laminar flow in the rest of the tube, which may not be the case - especially if you put one of those "trick" $99 cold air intakes on. The power gains you may or may not see may be due to bends in the plumbing causing the sensor area to be in a less-dense part of the total flow... so the ECU is tricked into running lean. (See? "more power" for a mere $99 . . .)
Just like an O2 sensor doesn't sense hydrocarbons, A MAF doesn't count oxygen molecules. It's an important point. All you really care about is how many free (not part of water) oxygen molecules are getting into the intake - not what temperature they are, the speed they are moving, how much the other gasses weigh, or what the humidity is.
If you could make a sensor that would accurately tell how many oxygen molecules are entering the intake (and maybe what temperature they are), real-time, You could eliminate a lot of other sensors and complexity. A/F ratio would be a simple real-time division: mL per moles O2.
There is the empirical observation that a MAP sensor costs about $50, and a modern MAF about $500. A speed-density based system is cheaper to produce (in quantity) after the fleet of tuners gets their job done. But I think it hinges on the flexibility of the engine, providing sufficient power and economy while adhering to emissions rules. Higher performance applications and supercharging may suggest more complexity is needed.
And as Bart commented, it's likely the engine is mapped and calibrated with highly sophisticated mass-airflow instruments gathering real-time data, and then your complex 50-layer maps can be overlayed/reduced to a simple speed-density-load map with input only from the production MAP sensor.