I think it's talking about a DC power distribution bus and basically it's saying that the source of the regulated voltage on that bus must have a very low output impedance. If it didn't have a low output impedance it wouldn't regulate very well.
The term "mask" just refers to the allowable "top-limit" of impedance that is acceptable and, as far as I can tell has no bearing on such concepts as impedance matching.
So, take the example of a 5V bus capable of putting out 100 watts. At low frequencies (0.01 Hz or less) the output impedance can be calculated from the graph as: -
0.002 x voltage\$^2\$ / power = 0.4 milli ohms
OK I found ECSS-E-ST-20C and it says (regarding this test): -
NOTE 2 Rationale for the impedance mask: It translates requirement
5.7.2i.1 of 1 % voltage change for 50 % load change in a domain of regulation up to 10 kHz bandwidth. In DC the integrator in the control
loop is designed to ensure no static error, in higher frequency,
between 10 kHz and 100 kHz it is likely that the inductance effect of
the components and connections are seen and the impedance rise not
always making feasible to respect the ideal impedance mask.
This proves to me it is about power sources feeding a power bus.