Incremental reading of static files in socket calculation #51
Description
The current read_results method depends on reading all chunks in the output file at once, which is very inefficient when the output file is huge (especially large MD and socket calculations).
It may need some substantial changes to the SparcBundle design, as all results are passed by raw_results, a possible solution is to allow read_raw_results to pass indices, and use ImageIterator to move along indices