Chunked particle data evaulation

[skuschel]

Particle Data may be carried out in chunks of a few million particles. This may have several advantages:

• If supported by the reader the data can be read chunk by chunk and therefore I would allow to process more particle than can fit into the computers memory. Requires heavy testing to ensure its working properly.
• Probably speeding up particle processing, as one chunk can be processed while another is read.
• The proposed change also provides the prerequisite to parallelize the particle data processing over several machines -- for example using ipyparallel.
• As this involves a possible restructuring of the project it has to be discussed before v1.0