Refactoring: Treat chunks/parts as normal jobs
Currently, a chunked download has just 1 job structure with a list of 'parts'. Thus queue_size() does not work correctly (returns just 1) and the parts are not downloaded in parallel (just 1 thread is started).
This is a regression from the last big refactoring.
A real fix needs to ad more fields to 'JOB', merging it effectively with the metalink structure. Not sure how to detect the last chunk downloaded to verify (checksum) the complete file.
If someone wants to work on this, please say so before you start.