Batch base Sync to improve large scale file async

[qiyuangong]

Description:
Currently, baseSync will fetch file list then insert diffs to metastore before processing file sync. This procedure works well when the number of files is not large. But, when there are billion of files in src dir, the processing time before file sync may be not acceptable.

Basic solution/idea (Async process file list):

1. Add all files to memory (very fast).
2. Batch insert diffs into metastore (slow due to metastore and network).
3. Processing file diffs when diffs are in metastore (fast).