There is no such thing as detectable "reversion" airflow to a MAF. Especially on a non turbocharged car. The MAF has no clue about the direction of flow, only how much energy it takes to keep the wire or film heated.
I doubt that the "negative" aspect of your graph is ever seen. Even at idle, you shoud see 2-5 g/s. I wonder why the graph is kg/h... that seems like ugly units for a MAF that needs to respond instantly (hopefully). Potatos, Apples. Too lazy to convert. My pedantry, not yours.
Simply put, if your car is at idle, it must be using air. Snap closure of a throttle butterfly at speed might cause some backflow, but again a MAF has no idea about the direction. So this is meaningless. Without knowing the BMW E39 Engine Control Unit (ECU) maps, I am certain 0% Throttle Position Sensor (TPS) is a cell (or cells) on the ECU map. Cells which ignore all MAF information, <1.7v or not. Your IDC (injector duty cycle) should go to zero (no fuel injection). No air, no fuel.
The graph implies a scaling, but you should not assume that every point on the graph is obtainable in running conditions.
And everything I just typed is meaningless without tacit understanding of what the ECU does with this information. So to answer "why should fuel management care" I would say it probably doesn't. I would also say (again) that there is no such thing... google how hot film or hot wire works... there is no element of any type MAF (wire/film) that I am familiar with that senses direction of airflow.
And actually, this is the crux of the problem. MAFs are awesome, but limited. In a turbo application, even more so. The ultimate goal is to find the proper Air Fuel Ratio. But even that is wrong... we imagine a utopian sensor that would count free oxygen molecules (O2), not NOx or water, that we can use for go-fast combustion. It should be termed Oxygen/Fuel Ratio.(Even "fuel" isn't right; are we talking 91 or 93 pump, E85 (flexfuel/alchol)?? Gotta count every oxygen and every carbon) On a nice fast ride (actually anything), such a sensor doesn't exist--especially at 300-500 cfm. So whatever MAF technology the Bavarians in Munich decided to use, it's probably top-notch. I don't know where the graph came from, but "negative" airflow areas are not plausible. Show me a below 1.7v reading on your working MAF on your running E39 and I'll eat my hat. Yours too.
Not something to be concerned about.
On edit: Forget every single thing you just read. I am totally, utterly, completely wrong.
Eine elektronische Hybridschaltung wertet diese Messdaten aus und ermöglicht so die genaue Erfassung der durchströmten Luftmenge einschließlich der Strömungsrichtung. ... A hybrid electronic circuit evaluates the measurement data and enables the accurate detection of the air volume, including the direction of flow.
This is from Bosch MAF HFM5
If you look at the schematic you can see that the output signal operational amplifier is biased by multiple temperature-sensing resistors, in different places on the film. So it knows which part of the film cools first, in a way. It can sense the direction of slightly reversed airflow.
I've got hat sauce to pick out. As far as the "why", I can only speculate (gee, great track record so far Steve) that this MAF behavior might be useful in a turbocharged application. After eating much crow, I still stand by my assertion that there will be little to no reversion airflow on a Normally Aspirated engine.
On edit edit: A better [speculation] might be now you have capability for a closed loop electronic Idle Air Controller, able to deal with a wide range of operating conditions, yielding potentially lower idle emissions, even on an Open Loop cold start--which has traditionally been a very difficult task to do well.