I think I disagree with your analysis of the "non-predicatability" of evolution (if I understand it right).
But first to answer your question about "other examples", the only truly "non-predictive" theory I can think of is quantum theory. The Heisenberg Uncertainty Principle even provides an equation that describes that fundamental non-predictability.
And a second case *might* be the theory of consciousness ... i.e. thoughts may be fundamentally unpredictable because they are affected by quantum events in the brain (read Roger Penrose).
However, evolution is different. Evolution falls in the same category as your snowflake example. Yes, the creation of a snowflake is an EVENT that can be predicted using, as you say, the intersection of many theories that individually make predictions ... but these individual predictions are *statistical* predictions. This is why, while the EVENT of the creation of a snowflake is predictable given certain conditions, the specific structure of the resulting snowflake is NOT predictable in any practical way. The same with the creation of a hurricane, or a storm on Jupiter.
The same is true of evolution. The EVENT of evolution by natural selection (more an ongoing PROCESS of many events) is indeed predictable, and is the intersection of many theories that individually make predictions (like genetics, and population dynamics). But again, these are *statistical* predictions. Thus, while we can predict that evolution WILL occur in certain conditions, we cannot predict the precise outcome ... the resulting organism.
In other words, the evolution of a specific organism is no more and no less "predictable" than the creation of a specific snowflake, or a specific hurricane.
All of these fall under the general category called "dynamic systems" ... otherwise known as chaos theory.
However, I also have to add one more comment ... as I detect a cynical attempt to diminish the theory of evolution as somehow different from all other theories, or even to imply that evolution should not qualify as a "theory" at all because it doesn't make predictions. If so, then you would be wrong ... evolution DOES make predictions (such as the locations and structures of what fossils will be found, the nature of genetic relationships between species, geographical distributions of species, molecular patterns in the DNA, etc. etc.). But making predictions is NOT the same as the theory being "predictable." Just because the theory cannot predict *specific* future outcomes doesn't make it unscientific. Evolution's inability to predict specific future organisms doesn't diminish it any more than meteorology's inability to predict a specific storm, or seismology's inability to predict a specific earthquake, or, for that mattter, quantum theory's inability to predict the decay of a specific atom. If that is what you mean by "non-predictability", then this is more the rule than the exception in science.
Nice try though.
{further edit}
Evolution absolutely *can* "retrospectively predict" a certain outcome. If we discover in colubine monkeys, two near-identical genes for two similar pancreatic enzymes, we can "retrospectively predict" that this is the result of a gene duplication event, followed by selective advantage for those monkeys who were better able to digest certain types of leaves that were their primary food in a certain environment. (See source.) I.e. genetics and endochrinology can "retrospectively predict" that specific enzyme in the same way that the chemistry of water, and crystallography can "retrospectively predict" a particular snowflake.
Why would one population A of cyanobacteria remain unchanged, while population B of cyanobacteria undergo evolutionary change? Because the members of B undergo different environmental pressures ... a population bottleneck, isolation in a tide pool, congregation near an oceanic vent, discovery of some new nutrient, or just some random new mutation, that through accidents of geography does not affect population A (cyanobacteria populations are worldwide, and subpopulations get isolated all the time). The result: two new strains A and B of cyanobacteria. Lather, rinse, repeat, for a million years, and you have many different strains, all undergoing different evolutionary pressures, and with different mutational input. Meanwhile, strain A continues to thrive quite happily in that environment in which it was originally optimal. Lather, rinse repeat, and A continues to thrive (as long as that environment continues to exist), while millions of different branches are each exploring different evolutionary niches ... some become fungi, some plants, some become posters on YA.