Incremental processing

[lfcnassif]

Make it possible to run any processing module (hash, signature, container expansion, carving, ocr, indexing...) after initial processing. This will also allow to resume processing with errors.

Probably the index will be broken into 2 (metadata and text indexes), so metadata index with processing flags will be easily and efficiently updated. Will break back compatibility, so is scheduled for v4.0.