How common is "decades ahead of his time" in real history?

Despite what you noted in your question, I would still argue that Leonardo Da Vinci meets the conditions you state for a scientist "ahead of his time."

He made a number of inventions during his life, including the parachute, functional scuba gear, and a primitive tank/armored car.

As for his flying machine designs, ornithopters very much like what Da Vinci designed are a real modern heavier-than-air flight technology. It is true that ornithopters are not the dominant mode of human flight today, so they are not very widely known, but they do work. The first successful manned ornithopter flight took place in 1942, and the first successful human-powered ornithopter flight took place in 2010.

While Da Vinci did not invent the idea of an ornithopter, he was the first to recognize that humans were not strong or agile enough to fly by simply strapping wings onto their arms. He was also the first to put serious effort into trying to overcome this problem. It should also be noted that his ornithopter design is a perfectly functional glider. While I am not aware of any tests, with a few minor modifications, it should be able to fly unmanned using a modern engine.

Enest Duchesne wasn't a scientist, and he made only a single discovery that went unnoticed for decades due to snobbish attitudes of French academia of the times, so I'm not sure whether he fits your criteria. Briefly, Duchesne discovered penicillin 32 years before Fleming, and used it to cure a case of typhoid. Unfortunately, he was deemed too young and unknown for anybody to pay any attention to his report.

Another good candidate is Reverend Thomas Bayes, who first correctly formulated how to use probability theory inductively, using what is now called Bayes' Theorem. The importance of this discovery was not fully understood until long after his death in 1761.

Bayes never published his discovery during his lifetime, but two years after his death his friend Richard Price found it in his notes in 1763 and published it as "An Essay towards solving a Problem in the Doctrine of Chances", which went almost entirely unheralded.

Bayes' Theorem was independently discovered a few years later in 1774 by the mathematician Pierre-Simon Laplace, who later learned of Thomas Bayes' earlier discovery in 1781.

After Laplace's death, Bayesian reasoning almost entirely disappeared from academic circles for over 100 years (though it still saw some infrequent use in industry and the military), until Harold Jeffreys published Theory of Probability in 1939.

Alan Turing was familiar with this work, and used Bayesian probability theory as part of the effort to decode Enigma codes during World War II. This, of course, was not public knowledge until many years later, since the British government tried to keep their code-breaking efforts secret.

Finally, in the 1950s, with the help of Alan Turing's former assistant I. J. Good and a few other characters, the power of Bayesian probability theory became widely recognized.

TL;DR:

Bayes' Theorem was discover by Thomas Bayes during his lifetime, but wasn't published until two years after his death in 1763. Except for a brief period of use by the mathematician Laplace, his work went largely unheeded until after World War II. It is now one of the two major interpretations of statistical reasoning.

The history of Bayes' Theorem is long and fascinating. An longer and better history of Bayes' Theorem can be found here.

Probably the best illustration of this particular concept is Friar Gregor Mendel.

Around 1900 biologists Hugo de Vries and Carl Correns were performing experiments trying to suss out the nature of inheritance, and found they had repeatable results they could not explain, and that did not fit in at all with the prevailing theories. Knowing they'd be torn a new one by supporters of those theories if they just released their results raw, they went looking for prior research showing similar results.

What they discovered was that this Moravian Augustinian Friar had not only produced the same results in experiments with peas 40 years earlier, but he'd come up with a workable consistent pair of theories that explained the results. In the intervening 40 years (during which he passed away) his work had been cited all of 3 times, but otherwise essentially forgotten. His two theories we now call Mendel's Laws of Inheritance, and Mendel himself is now considered the father of modern genetics.

As for the general case, I'd argue that this kind of person probably exists in every society and age, but for it to be a true advance you generally have to look to societies advanced enough to give the person a good starting base, and its probably a far more typical result for a Mendel's work to never be rediscovered later. So having the rare such genius, and the good fortune to not lose their work, is a total stroke of luck when it happens.

Srinivasa Ramanujan is a good candidate, based on your criterion of "their work has to be independently proven valid at a much later date." Ramanujan often chose not to prove his nearly 3,900 results, many of which were unconventional for the time. In the decades after his death in 1920, mathematicians would prove him correct again and again and again. Some results were incorrect, but Ramanujan's record is still remarkable.