Decide how to handle writing unicode strings to HDF5

[anjakefala]

memh5 and andata were updated to automatically convert strings to unicode. This is great if you stay within the memh5 framework for writing and reading data to HDF5, but can cause problems if you try to write directly with h5py. h5py does not handle the writing of unicode strings to HDF5 files.

For future analyzers, we should consider populating a memh5.BasicCont (instead of the h5py file handler) with the various datasets, and then use the save method of that container to write to disk. That way it would do all the unicode string conversions automatically.

[anjakefala]

Relevant #150