coin_scraper.py should be split in several independent processes

1. GET /tasks and find the new tasks.
2. GET /taskDetail, taskWorkItems for all tasks that are ready.
3. GET /workItems in a loop for the workItems that are still running.
4. Download, parse, process and store testresults, for completed workitems.

These tasks are independent and can each one repeat itself in an endless loop in its own process.