dplyr functions should not shadow sparkR functions

[piccolbo]

collect count are examples. On the other hand, I see these two packages used with plyrmr in a pipeline that starts distributed and ends local. Needs to think about naming madness.