Integrate the new Mercurial loader into the tasks system

The new Mercurial loader is ready in its incremental form.

The tasks system should differentiate between mercurial and mercurial_disk, and allow the CLI/scheduler to pass both directory and URLs. This is important for the BitBucket Mercurial Archive since the loading will be done from disk as a oneshot, but should list the original URL in the metadata.

This is very likely to be a matter of minutes of work in the tasks.py and tasks_from_disk.py, but it's probably better if someone from the SWH code team either tells me what to do or does the work themselves.

---

Migrated from T3335 (view on Phabricator)