It is frustrating beyond belief to get the same champ or same class AG repeatedly. But that doesn’t prove anything is rigged—in fact, “clumpiness” seems to be a feature of the particular methodology used in MCoC (and a lot of other RNGs).

It is a feature of pRNGs only in the sense that it is a feature of randomness in general. You would expect random numbers to exhibit a certain degree of "clumpiness" and any RNG that doesn't would be considered broken.

A big problem with randomness and humans is humans have a conception of what "random" is that is completely wrong. They think randomness is more "even" than it really is, that it ought to be as "featureless" as possible. So much so there's an infamous story where a college professor asks students to generate their own list of random numbers and then shuffle those into a set of actual random numbers, and the teacher can generally pick out the ones made by students almost every time. Human generated random numbers are, for lack of a better way of putting it, always trying too hard to look random. Actual random numbers, ironically, don't look random.

RNGs aren't "designed" to do anything specific. They don't "target" some "kind" of randomness. Rather, pRNGS are designed to avoid non-random behavior. They are designed to not generate patterns, except at the statistical rate those patterns ought to appear in. For the most part, when it comes to frequency (how often each number appears) and correlation (how often two numbers appear together, or are duplicated, or similar) all pRNGS tend to look alike most of the time. They kind of have to, because while there is an infinite number of ways to be non-random, there's only one way to be random. Random numbers do not have species.