So the "executable grammars" are strongly related to the idea of "executable papers" in the sciences.
Another aspect is that any stats quoted (e.g. the number of nouns of a particular inflectional class in a particular corpus) can be verified at any point.
Much like a framework that allows code fragments in software documentation to be tested to make sure they run properly, one aspect of this "executable grammar" idea is that any references to language examples should be linked back to a corpus in such a way that it can be verified that both the example and citation are correct.
In many ways, a better way of thinking about what I mean is applying concepts like automated testing, doc tests and literate programming to descriptive natural language grammars.
I'm not talking about grammars in the formal language sense but in the sense of a descriptive grammar or reference grammar in book form.
That said, the original generative goal of a system of rules by which grammaticality can be tests is somewhat aligned with what I'm talking about.
In a couple of weeks, I'll be presenting to the BibleTech 2015 conference on Better Greek Learning Through Better Greek Databases. In November, I'll be presenting at the Society of Biblical Literature's Annual Meeting on my Morphological Lexicon of New Testament Greek.
One concept I'll be touching on in both presentations is the notation of "executable grammars" of Greek (or any other language).