Now 'classical systems' are systems made of so many elements that we can only make statistical statements about it because we can't ever determine the state of each of it's constituents.
I suspect for some time now that this is also the source for the "quantum randomness". In principle, everything is deterministic, but to observe something we always have interactions with a "big classical system" (in the end it's always our brain). And because we don't know the exact state of those systems but only statistical averages, quantum mechanics looks random for us, even if it's perfectly deterministic in itself. But that's my personal view of the matter.
One of the postulates of QM is that after a measurement the system is in a well-defined state (the one we measured). But this is aphysical.
It's an artifact of the deliberate choice to formally model measurement in a way that simultaneously recognizes that all measurements disturb the measured system, while also wanting to abstract away the particular hardware used.
So, I'm starting to see wavefunction collapse and the whole Copenhagen interpretation as artifacts of the the formalism, not any kind of physical truth.
One way to do a partial measurement is to make the conditional effects less orthogonal to each other, e.g. cause a target qubit to conditionally rotate by 22 degrees instead of a full 180 degrees.
Another possible concept for a partial measurement is not distinguishing between all of the cases. For example, if a system can be in the x=-1, x=0, or x=1 cases then it's possible to do a measurement that distinguishes x=-1 from x in {0,1} without collapsing all the way down to x=0 or x=1.
So anything related to light then as well?
Same thing goes for many things, for example letting some hydrogen gas out of a gas bottle: The gas will never go back into the bottle by itself (even if there is a very very small probability that this may happen, in practice it never will). Because the behavior of this kind of systems is described in thermodynamics those systems are called "thermodynamically irreversible".
In the end it simply means that its much more easy to break things than to assemble them again (which also leads to the concept of the always increasing entropy).
It does NOT mean a human. It can interact with anything, another particle even.
However, not all interactions with another particle count as an observation, and I think the poster asked for a more precise description of the boundary, and why it would not be a boundary but actually appears to be a continuum.
The general public does, most explanations of the double slit experiment you can find lead people to believe that mind affects reality because the word observation confuses people into thinking consciousnesses is involved.