Has the speed of light ever been measured in vacuum? According to: https://en.wikipedia.org/wiki/Cosmic_microwave_background the CMB (Cosmic Microwave Background) "is faint cosmic background radiation filling all space"
Also, https://en.wikipedia.org/wiki/Vacuum#Outer_space says "no vacuum is truly perfect, not even in interstellar space, where there are still a few hydrogen atoms per cubic meter"
And https://physics.stackexchange.com/tags/vacuum/info: "This rather theoretical requirement is never achieved in practice, because even if space does not contain any atoms / electrons / nucleons, it does contain a lot of photons and neutrinos. But we still call it a vacuum, as an approximation of the theoretical vacuum."
Then on https://en.wikipedia.org/wiki/Speed_of_light we have "The speed of light in vacuum, commonly denoted c, is a universal physical constant important in many areas of physics. Its exact value is 299,792,458 metres per second (approximately 300,000 km/s (186,000 mi/s)). It is exact because by international agreement a metre is defined as the length of the path travelled by light in vacuum during a time interval of 1/299792458 second"
Given that, at least according to Wikipedia, in practice vacuum does not exist, and it's filled with radiation, is it actually possible to measure the speed of light in vacuum? Also, has anyone even observed light in vacuum, ever?
If not, what types of vacuums has light been measured in? Are there any records about this?
Additionally, if true vacuum doesn't even seem to exist and it's filled with radiation (CMB), can we really assume that light doesn't need a medium to propagate? Wouldn't it effectively be propagating through whatever is filling up the space? Would we ever be able to tell the difference?
 A: Trying to address your confusion about whether the definitions are circular...
I think the root of your confusion may be in understanding what this means from the Wikipedia article you quoted:

Its exact value is 299,792,458 metres per second

You bolded the above sentence, but it is not really the important one. The important one was later in your quote:

a metre is defined as the length of the path travelled by light in vacuum during a time interval of 1/299792458 second 

I think you also may be confused about the difference between values and the physical quantities they represent. 
So to determine what a meter is based on the above definition, you first need to measure the speed of light. Imagine you are able to make an experiment that measures how far light goes in 1 second. Perhaps it draws pencil marks on two walls very far apart. Say you run the experiment 10 times and look at the different results. The outputs from the different runs of the are not values like [299792458.1 m, 299792457.7 m, 299792458.3 m, ...]. Instead, the outputs are all called "299792458 m". Of course they do have measurement error, but the error is in the actual physical measurement (the locations of the pencil marks).
If someone wants to use this experiment to, say, manufacture a very accurate meter stick (maybe a "light-second stick" in this case), they would need to actually use their meter stick to measure the distance between the pencil marks on the wall, and use that to know whether their meter stick is too large or two small and adjust their manufacturing process based on that.
Does this give you any insight, or am I off base here about your confusion?
A: Science is full of ideals in its wordings.  This is one of them.
SI has fixed the speed of light in a vacuum to be 299,792,458 m/s.  If there was indeed light propagating through a perfect vacuum, that would be its speed ... because we define it to be.
For practical purposes, however, we need to be able to design experiments with which to measure distances using this definition.  We have done these sorts of experiment regularly in high vacuum, on par with or more extreme than the vacuum of interstellar space.  When we look at the effect matter has in slowing the speed of light, we find that the difference between its speed in a perfect vacuum and an achievable vacuum is smaller than the measurement error on our experimental devices.  Before we had fixed the speed of light to be a constant, we had measured it to within 1 m/s.
How much of an effect does it have?  I'm having trouble finding sources to give a definitive answer, but based on the refractive index of hydrogen as a function of pressure, I would expect interstellar levels of hydrogen to slow light by a factor on the order of µm/s.  It's very difficult to measure physical things to 8 or 9 digits, and µm/s is 15 digits away from the speed of light, so our measurements in a high vacuum are as usable as if they were in a perfect vacuum.
If, at some point in the future, we discover that this approach is flawed, we will amend it, as has been done several times before—the most recent amendment being fixing the kilogram as a function of several fundamental constants.
A: The “few atoms of hydrogen per cubic meter” is a vacuum for visible light. The adjective “perfect” is both unnecessary and irrelevant in describing light propagating through a vacuum. 
The wavelength of visible light is around 500 nm. So roughly $8 \ 10^{18}$ wavelengths fit in a cubic meter. A few hydrogen atoms are utterly irrelevant. Their presence cannot explain the wave behavior of light and from the perspective of such a phenomena the region is vacuum. 
Since neutrinos don’t interact with light, their presence is also unimportant, regardless of their quantity. And photons are the same thing as light so it hardly makes sense to complain about the presence of photons when discussing the presence of light. 
A: 
the CMB (Cosmic Microwave Background) "is faint cosmic background radiation filling all space" [...]
Given that, at least according to Wikipedia, in practice vacuum does not exist, and it's filled with radiation, [...]

You seem to be assuming that a vacuum in which electromagnetic radiation (such as the CMB) is present is not a “true” vacuum. But if you define the vacuum in such a way, then we could never measure the speed of light in a vacuum, because as soon as you inject light into the vacuum it stops being a perfect vacuum. See the problem here? Your notion of what a vacuum is is a much too purist and restrictive one.

Additionally, if true vacuum doesn't even seem to exist and it's filled with radiation (CMB), can we really assume that light doesn't need a medium to propagate?

I think what you mean is that light can propagate in a vacuum (that is, that light doesn’t need matter to propagate), not that light doesn’t need a medium to propagate. As for a “medium”, the truth is we don’t really know what light is and how or why it propagates, so the statement about a medium belongs to the realms of philosophy or metaphysics, not physics. We do have excellent mathematical descriptions of light, the vacuum, and other physical phenomena. Our mathematical models say light propagates in a vacuum, and all existing experimental evidence confirms this to an overwhelming degree of certainty. That’s really all that physics has to say about this issue of propagation.
A: There are many answers which point at different ways of understanding the question and convey different types of information about it. I am very thankful for everyone that has contributed to this discussion.
This answer is to condense the info contained in the other answers so far, which together I feel do answer the posted question.
Here's what I learned:
1) We have actually never observed light in "true vacuum", as "true vacuum" doesn't really exist (or it is ill-defined)
2) Given 1), we haven't really been able to measure the speed of light in vacuum. However, this doesn't seem to matter, as we have agreed to define the speed of light as a fixed quantity
3) There is no universally-accepted definition of vacuum, it means different things in different fields or contexts
4) Despite 1), 2), and 3), it doesn't really matter for the purposes of physics, because the models work anyway and the models are not reality, just useful mathematical tools to help us make predictions
5) There aren't really any absolute units of anything. Constants and units are abstract concepts defined through the mathematical models of physics and the values of the units are defined by agreement and governed by the SI. So for example, there is no problem with defining the speed of light as a function of the meter and at the same time defining the meter in terms of the speed of light
