Does observation in quantum theories always imply interaction (affecting quantum system with photons, electromagnetic fields, etc.)? The term observation is obscure. 
As I see so far, observation is always done by means of affecting (!) the quantum system by some means - often photons or electromagnetic waves or whatever else.
For example reading Wikipedia about Zeno effect - there measurement is done by applying a sequence of ultraviolet pulses. Further wikipedia writes: "As expected, the ultraviolet pulses suppressed the evolution of the system into the excited state." So it is just those ultraviolet pulses that did the slow-down effect!
So why quantum physics so much emphasizes those "observation paradoxes" (quantum system changes or becomes different only when being observed) - it is clear that something changes (or becomes different) when there is interaction/influence (I mean observation always means interaction/influence, no observation then means no such interaction/influence). I understand you may counter saying that until observation there is nothing manifested (just a superposition), but often it is said that "observation changes the quantum state", right? Isn't that stupid - obviously any interaction may cause changes?
In my Zeno example why do physicists speak of just observation, not the effects of interaction of their (quantum) system with ultraviolet waves of certain frequency, amplitude, etc?  Or the same effect would be observed if we use laser (photons) or accelerated protons instead of ultraviolet? 
But even then we cannot generalize that "any observation" (= affecting the quantum system with any means) would produce the same (Zeno) effect! 
However it seems that physists do make this generalization somehow, because they say that "observation slows-down transition of particles from higher energy level to lower energy layer". They don't say "this interaction and that interaction during measurement produces this effect".
 A: 
Does observation in quantum theories always imply interaction...?

Observation always implies interaction period, whether in quantum theory or any other theory. We cannot observe something unless that thing influences its surroundings in some way (scattering light, influencing the motion of nearby objects, etc). This is not specific to quantum theory.

As I see so far, observation is always done by means of affecting (!) the quantum system by some means...

In principle, it's the other way around: observation requires that the system of interest affects its surroundings (e.g., us) in some way. However, according to the action principle (from which conservation laws are derived), influences must go both ways — so in order for the thing being observed to influence its surroundings, its surroundings must also influence the thing being observed.

So why quantum physics so much emphasizes those "observation paradoxes" ...

While observation requires interaction in any physical theory, the role of observation in quantum theory has a new twist. In quantum theory, the interactions associated with observation (I mean the physical processes, whether or not any people are involved) result in a state that is typically a superposition of two or more terms that cannot be mixed with each other by the mathematical operators representing any feasible future measurements. In other words, we might as well end up with a state having only one of those terms, but (1) quantum theory can't predict which one, and (2) quantum theory can't even predict how often each one will occur when the experiment is repeated, except with the help of an additional postulate specifically dedicated to that role: Born's rule.$^{[1]}$ All of the various "observation paradoxes" (at least the legitimate ones) have their roots in this. 
Becoming familiar with the CHSH inequality, and the fact that it is violated in the real world (as are other Bell inequalitites), is a good way to gain some appreciation for the non-trivial role of observation/measurement in nature, independently of the formalism of quantum theory.

it seems that physists... say that "observation slows-down transition of particles from higher energy level to lower energy layer".

Whether or not a given pattern of observation/measurement leads to the Zeno effect depends on several things. Observation is a physical process that (in principle) can be described in quantum theory just like any other physical process, and only after obtaining a state of the form described above can we safely project onto one of those terms with a frequency given by Born's rule. It is certainly not true that "any observation" would produce the same Zeno effect.

Footnotes about Born's rule:
$^{[1]}$ This is what I mean by "Born's rule": If $|\psi\rangle$ is the post-measurement state, which may be a superposition of two or more practically-unmixable terms, and if $P$ is the projection operator onto one of the practically-unmixable terms, then the relative frequency of empirically getting the result $P|\psi\rangle$ is $0\leq \langle\psi|P|\psi\rangle/\langle\psi|\psi\rangle\leq 1$. 
The Mackey-Gleason theorem shows that Born's rule is essentially the unique rule that is compatible with the rest of quantum theory's structure. However, that theorem does not constitute a derivation of Born's rule from the rest of quantum theory's structure, for a simple reason: the Mackey-Gleason theorem is not sensitive to whether or not the observable in question has actually been measured — that is, to whether or not the state is a superposition of practically-unmixable terms that are eigenstates of the observable in question. That's why Born's rule is described as an "additional postulate specifically dedicated to that role."
A: In elementary particle physics interactions means an exchange of four momenta between incoming and out going particles, where observable are just the incoming and outgoing particles , and the effect of the four momentum exchange is measured by using energy momentum and angular momentum conservation, and a number of quantum number conservations. In this field observation and interaction is the same. One observes the effects of interaction. 
But physics is not just elementary particle physics. It is concerned with complex systems, like protons and neutrons, nuclei, atoms, lattices etc. At the level of nuclei for example, there are continuous interactions between the protons and the neutrons, which are not observable directly. Levels of mathematics have to be used, so in this sense observations are a subset of interactions.

In my Zenon example why do physicists speak of just observation, not the effects of interaction of their (quantum) system with ultraviolet waves of certain frequency, amplitude, etc

Because the "interaction" is not one of the fundamental ones, but a complicated mathematically synergy of UV photons gives a behavior that is emergent from the fundamental electromagnetic interaction which involves one photon at a time. It is a great multiplicity of fundamental interactions, and thus they stick  to "observations".
Simple quantum mechanical models have been developed for complex systems as for example the phonons , and then one might have a simple interaction model and be back in a framework where interaction and observation can be the sides of the same coin.
