First diff showcasing the work done so far in order to ingest into the archive
the numerous source packages hosted on the npm registry.
This work is greatly inspired from the PyPI loader written by @ardumont.
Apart some slight differences between PyPI and npm, the process to load
their content into the archive is barely the same: read package metatada
from a JSON endpoint then download tarballs and load them into the archive.
This clearly calls for a future refactoring to remove some code duplication
and establish a loading worflow to ingest content from other package managers.
The loader can be tested through the following command line:
```
$ python3 -m swh.loader.npm.loader <npm_package_name>
```
To correctly browse the results with `swh-web`, you will need to use its
development version as current package published on PyPI does not handle
browsing this new type of origin yet.
Related T1379