1 read one file from dataset and convert the file into IPLDs and save it into blockstore
2 records the meta data of the file in the dataset, including: relative path in the dataset, filename, file size,file cid
3 make the working process can be paused or restarted at file level
4 could check if the import work completed