That's actually not the definition used in biology. Simple example: The entire genus Amphiprion (aka anemonefish) are speciated due to appearance and regional locality. They can and do interbreed when introduced into each others populations, and have fertile offspring that are mixed species. They are declared species due to their appearance, and their natural range's prevention of interbreeding. That said, there are areas where some ranges overlap and it's not uncommon that one species mates with another and has hybrid offspring.

Being genetically capable of interbreeding makes sense as a definition of speciation, but that is not how species are currently identified. Natural range, aesthetics, and more than that, whether or not they DO interbreed in the wild (vs physically/genetically capable) play a more significant role in determining species.

There are many 'hybrid' species in nature, which is what I feel warrants some revision, perhaps using genetic data to restructure many classifications as subspecies rather than independent species.

edit: Sorry, I think I may have misinterpreted what you said to some extent and think we generally agree. I will maintain that according to the classification methods described, people such as the Sentinelese, may well be considered a separate species to an unbiased observer using the same guidelines we apply to other animals due to slightly different physical characteristics and social/geographical reproductive isolation prevents the possibility of breeding with anyone outside of their group.