Dear lurscher, the quote is the kind of C-physics described by the C-word which is a favorite word of mine but is discouraged on this server, so I won't use it - but you have used it. You don't misunderstand anything - quite on the contrary, you're right on the money.
These comments about a non-existent test of parity in the equivalence principle are due to "Uncle Al" Schwartz, see e.g. the first comment under
http://www.iac.es/galeria/masc/Outreach_files/Dark-energy_particle_spotted_NatureNews290609.pdf
to see that even extremely microscopic details about the "idea" coincide with the paragraph you quoted. "Uncle Al" is known to most physics bloggers - as well as many other physics forums.
From a physics viewpoint, the claims are completely preposterous. First of all, one needs pretty special objects to get a parity-violating physics in 4 spacetime dimensions: either chiral (Weyl) spinors or self-dual 2-forms (and their interactions). None of those things is included in GR; Einstein-Cartan theory; conventional affine, teleparallel, and noncommutative gravitational theories - which is why all these gravitational theories automatically preserve parity. Moreover, self-dual 2-forms don't exist in 3+1 dimensions (although they do exist in 4+0 dimensions) because $*^2=-1$ in 3+1 dimensions.
As you correctly say, only the weak interactions as expressed by the electroweak theory violate parity which is because the gauge field interacts with chiral (Weyl) spinors. The gauge fields $A_\mu$, when coupled to spinors, get naturally multiplied by $\gamma^\mu$ which switches the chirality (because it anticommutes with $\gamma_5$); on the other hand, $g_{\mu\nu}$ with an even number of indices preserves the chirality, so the spinors interactions with the metric have to be parity-preserving.
Moreover, the existing tests of the equivalence principle de facto eliminate the possibility of substantially different interactions of left-handed and right-handed helixes, too. For example, Newton's apple contains vitamin C (L-ascorbic acid) which is parity-asymmetric, and it would accelerate differently if the laws of gravity cared about the mirror images. There's really no way compatible with an effective field theory how to make the two mirror objects accelerate differently.
For some semi-serious paper by authors from well-known places who talk about gravitational parity violation, see e.g.
http://arxiv.org/abs/arXiv:1005.3310
and the papers mentioned by it (by Stephon Alexander and others). It would still be impossible for me to say that any scenario in those papers is defensible but at least those papers don't sound as a self-evident nonsense of the kind that "Uncle Al" Schwartz has managed to incorporate into the Wikipedia page.
