Draft: block_log.index refactored to block_log.artifacts file, since it stores more informations than block_log file offsets

Artifacts generation is performed in multithreaded way, to limit time needed for it. Supports also generation resume in case existing file is points to lower block data than block_log file. Artifacts file additionally stores dedicated header, what allows semantic verification of block_log and block_log.artifacts files (this must be still implemneted). Additionally, block_log file walker has been implemented to be used for general purpose block_log processing. Now it's integrated directly into block_log class, but in the future we can decide about creating an external tool class for it, to avoid using whole block_log interface.