I don't know if this is useful to anyone here, but Matt and I were able to add Chombo support to yt. It's still super alpha, but it works. Attached is a slice through a test dataset ran by PS Li here at Berkeley. The code is in the mercurial repository under the "chombo" branch. I hope to move it back to the main yt branch (in mercurial) within the next week or so.
Chombo's datafile format uses a single hdf5 datafile for all processors, levels, and grids, so it may be of some interest for parallel IO.