I think a repository of RecDescent grammars is an excellent idea. I don't think they should be included in the module distribution itself for fear of excessive bloat. A couple of alternatives:

1. Someone could volunteer to maintain a Parse::RecDescent::Grammar pseudo-module, that aggregates all the contributed grammars. There could even be a front-end analogous to Regexp::Common.
2. We could just use Regexp::Common directly, by providing a top-level $RE{grammar} namespace.
3. We could ask for a RECDESCENT (pseudo)ID on CPAN and use that directory to house RecDescent grammars.
4. Someone could set up a web-site that allowed people to submit and then update their grammars

I would support any of these approaches, though, unfortunately, I don't have the spare capacity to implement any of them myself. Personally, I favour the P::RD::G approach, since it puts the grammars on the CPAN, in the correct namespace, where search.cpan.org is likely to draw them to the attention of potential beneficiaries.