It's an awful language w/o formal grammar and plenty of the same problems people knew SQL has had for ages: modules, variety of collections other than the only main kind.
Writing in Cypher takes a lot of guessing, and there's no real explanation for why sometimes things that seem like they should work -- don't. It feels like a language designed by an amateur, built around a gimmick syntax of (x)-[y]->(z), but as soon as you need something a little more complex, there's nothing.
GP is in a completely different category of things. It has a deep mathematical formalism behind it, a unique and worth of exploring computation model.
Cypher in comparison to GP is like my 3 y.o doodles compared to Rembrandt's doodles. Superficially, both are a bunch of squiggly lines...
Was expecting to find the grammar, but found nothing of the kind.
From my experience with how Cypher is implemented in Neo4j (admittedly, from a while back), it was a bunch of regular expressions and some hand-tailored string processing.
There's nothing useful (for programmers who want to use Cypher) in the link you provided, just some information about the work group that works on this language. Without much research, I'd imagine that this is an organization sponsored or otherwise supported by Neo4j to promote the use of their database.
OpenCypher is a group that includes Microsoft and others. The results will likely be the standard.
Being able to match patterns in a graph is pretty much how you retrieve data. Being able to identify properties of relationships is what Cypher does extremely well.
The GP thing from the OP is mathematical and useless for most software engineers.
